Changelog#

v2.8.0 - 2026-03-16#

This release brings important improvements to the BaseRestartWorkChain, the engine, stashing, typing coverage, and dependency updates. It also includes several fixes for connection and resource leaks (with further improvements planned for upcoming releases).

Highlights:

Behavior changes: The following changes may affect existing workflows but do not constitute API breaking changes that would warrant a major version bump. See the Behavior changes section for details.

BaseRestartWorkChain enhancements#

Two new inputs have been added to BaseRestartWorkChain to give users more control over failure handling:

on_unhandled_failure (behavior change)

Previously, BaseRestartWorkChain would automatically restart once on unhandled failures. This behavior has changed: the default is now to abort immediately to save computational resources. To restore the previous behavior, explicitly set on_unhandled_failure='restart_once'.

The available options are:

  • abort (default): Abort immediately with ERROR_UNHANDLED_FAILURE

  • pause: Pause for user inspection

  • restart_once: Restart once, then abort (previous default behavior)

  • restart_and_pause: Restart once, then pause if it fails again

from aiida.engine import submit

inputs = {
    'on_unhandled_failure': 'restart_once',  # Restore v2.7 behavior
    # ... other inputs
}
submit(MyRestartWorkChain, **inputs)

pause_on_max_iterations

A new pause_on_max_iterations input allows the work chain to pause instead of aborting when the maximum iteration limit is reached. This is useful when you want to inspect the calculation state before deciding whether to continue or abort:

inputs = {
    'max_iterations': 3,
    'pause_on_max_iterations': True,
    # ... other inputs
}

When paused, use verdi process play <PK> to resume or verdi process kill <PK> to abort.

Replaced nest_asyncio with greenback#

The internal async event loop handling has been refactored to use greenback instead of nest_asyncio (#7206). This change brings several benefits:

  • Drops the deprecated nest_asyncio dependency (the package is no longer maintained)

  • Clearer tracebacks when debugging engine code

  • Ability to add breakpoints in engine modules during development

Important change for Jupyter notebook users: When using AiiDA in a Jupyter notebook, load_profile() must now be called in a separate cell before running engine processes:

# Cell 1: Load the profile
from aiida import load_profile
load_profile()

# Cell 2: Now you can run processes
from aiida.engine import run
from aiida.plugins import CalculationFactory
# ...

This is required because the greenback context is established after the cell containing load_profile() completes.

Unstashing support#

Building on the stashing feature introduced in v2.7.0, this release adds unstashing—the ability to retrieve previously stashed data back to the remote working directory (#6826).

The new UnstashCalculation allows you to restore stashed files before running a calculation that needs them:

from aiida.plugins import CalculationFactory
from aiida.engine import run
from aiida.orm import load_node

UnstashCalculation = CalculationFactory('core.unstash')

# Load the stashed data from a previous calculation
stash_node = load_node(<STASH_NODE_PK>)

inputs = {
    'stash': stash_node,
    'metadata': {
        'computer': stash_node.computer,
        'options': {
            'resources': {'num_machines': 1},
        },
    },
}

result = run(UnstashCalculation, **inputs)
# The unstashed files are now available in result['remote_folder']

Additionally, RemoteStashCustomData (#6777) enables stashing via custom scripts for specialized data handling needs.

TrajectoryData: periodic boundary conditions support#

The TrajectoryData class now properly stores and retrieves periodic boundary conditions (PBC) (#7079). Previously, when creating a trajectory from StructureData instances with specific PBC settings (e.g., periodic in x and y but not z), this information was lost and the structures retrieved from a TrajectoryData would always default to (True, True, True).

Now the periodic boundary conditions are stored in the attributes and can be retrieved using the .pbc property:

from aiida import orm

structure = orm.StructureData(
    cell=[[1.0, 0.0, 0.0], [0.0, 1.0, 0.0], [0.0, 0.0, 1.0]],
    pbc=(True, True, False)
)
structure.append_atom(position=(0.0, 0.0, 0.0), symbols='H')

trajectory = orm.TrajectoryData([structure])

trajectory.pbc
# (True, True, False)

Structures retrieved from the TrajectoryData also have the correct boundary conditions:

trajectory.get_step_structure(0).pbc
# (True, True, False)

Backward compatibility: For TrajectoryData created before v2.8, trajectory.pbc returns None.

Deprecation: Providing cells without explicit pbc to set_trajectory() now issues a deprecation warning and defaults to (True, True, True). This will raise a ValueError in v3.0.

Typing improvements#

This release significantly expands type annotation coverage across the codebase. The following modules now have strict typing enabled:

  • aiida.common

  • aiida.repository

  • aiida.schedulers

  • aiida.tools.graph

  • aiida.tools.query

  • aiida.cmdline.utils

  • aiida.cmdline.params

  • aiida.cmdline.commands

  • aiida.engine.processes

These improvements enable better IDE support, catch more bugs at development time, and make the codebase more maintainable. Plugin developers will benefit from improved autocompletion and type checking when using the AiiDA API.

Dependency updates#

numpy 2.x support (#6899): AiiDA now supports numpy 2.x, allowing use of the latest numpy features and performance improvements.

click 8.2 support (#6959): The CLI is now compatible with click 8.2, which includes various improvements to command-line parsing.

verdi process repair: orphaned database connection cleanup#

When daemon workers crash or are killed without proper cleanup, their PostgreSQL connections can remain open. These connections hold locks that can block other processes indefinitely, causing workchains and calcjobs to hang.

The verdi process repair command now detects and terminates these orphaned connections (#7207). Since the command requires the daemon to be stopped, any remaining PostgreSQL connections (other than the current session) are likely from crashed processes. A confirmation prompt is shown before terminating connections, as legitimate connections (e.g. from Jupyter notebooks) may also be affected. Use --force to skip the prompt.

Deprecation of core.ssh transport#

The core.ssh transport plugin is now deprecated in favor of core.ssh_async (#7175). The async transport, introduced in v2.7.0, offers better performance for concurrent operations and improved configuration through ~/.ssh/config.

When configuring new computers, you will see a deprecation warning if selecting core.ssh. We recommend migrating existing computers to core.ssh_async.

This release also adds 2FA support for core.ssh_async with improved authentication script handling (#7184).

Behavior changes#

StructureData.cell_angles: returns None instead of nan (#6820)

The StructureData.cell_angles property now handles zero-length cell vectors differently:

  • Returns None for angles that cannot be computed due to zero-length vectors (previously returned nan)

  • Raises ValueError when all cell vectors have zero length

  • Handles numerical edge cases by clamping cosine values to [-1, 1], preventing nan results for nearly-parallel vectors

If your code checks for nan values in cell_angles, update it to check for None instead:

# Before (v2.7 and earlier)
angles = structure.cell_angles
if any(math.isnan(a) for a in angles):
    # handle undefined angles

# After (v2.8+)
angles = structure.cell_angles
if any(a is None for a in angles):
    # handle undefined angles

Portable codes: job scripts now include ./ prefix (#7080)

Portable code executables located in the top-level directory are now invoked with a ./ prefix (e.g., ./mycode.py instead of mycode.py). This fixes execution on systems where the current directory is not in PATH. If you parse job scripts, be aware that the executable line will look different.

Stashing: rejects RemoteData from still-running calculations (#7147)

StashCalculation now validates that the source RemoteData node comes from a sealed (finished) calculation. Attempting to stash from a still-running calculation will raise an error, preventing incomplete stashes. Use the direct stashing method (metadata.options.stash) or wrap StashCalculation in a workflow to ensure proper sequencing.

core.ssh_async: glob now returns empty list instead of raising error (#6950)

The glob method of core.ssh_async transport now returns an empty list when no files match the pattern, instead of raising OSError. This aligns the behavior with core.ssh transport.

Transport: --safe-interval default changed from 30s to 15s (#6943)

The default value for --safe-interval (connection cooldown time) has been reduced from 30 seconds to 15 seconds. Additionally, this option now has a non-interactive default, so scripted computer setup (verdi computer configure) will no longer hang waiting for input.

SQLite storage: database lock warning downgraded to INFO level (#7121)

The warning about failed writes due to SQLite database locks (common with core.sqlite_dos storage) has been downgraded from WARNING to INFO level. This reduces log noise for SQLite-based profiles where these transient locks are expected and harmless.

Improved ProcessBuilder representation (#7285, #7290)

The ProcessBuilder now has a proper __str__ method, so its clean YAML-like representation is available in all Python environments via print(), not just IPython and Jupyter. The existing _repr_pretty_ method (used by IPython/Jupyter) has also been improved to prune empty namespaces (e.g. stash, metadata.options) from the output, reducing noise when inspecting builder contents. Note that this affects the print() output for a ProcessBuilder:

from aiida.calculations.arithmetic.add import ArithmeticAddCalculation
builder = ArithmeticAddCalculation.get_builder()
print(builder)

# Process class: ArithmeticAddCalculation
# Inputs:
# {}

In case the old string representation is still desired, you can wrap the builder in a dict:

print(dict(builder))

# {'metadata': {'options': {'unstash': {}, 'stash': {}}}, 'monitors': {}}

Features#

Engine#

  • ProcessBuilder: use __str__ instead of __repr__ (#7290) [4311f04c9]

  • ProcessBuilder: add __repr__ (#7285) [c79d22a5f]

  • ProcessBuilder: prune empty namespaces in _repr_pretty (#7285) [3be99a61a]

  • Add timestamp and PID to worker log (#7271) [dacaa47c5]

  • Revert workaround of increasing maximum stack size (#7230) [4c6bba90c]

  • Replace nest_asyncio with greenback support (#7206) [2c010c6dc]

  • BaseRestartWorkChain: add pause_on_max_iterations input (#7139) [24e840cd0]

  • Asyncify request_transport context manager (#7209) [41acc86c6]

  • BaseRestartWorkChain: add on_unhandled_failure input (#7116) [a770d7a6e]

Stashing#

  • Add fail_on_missing option to control stashing behavior for missing files (#7140) [c10fef19c]

  • Add warning when deleting StashCalculation nodes with --clean-workdir (#7187) [381d3932b]

  • Add validation to reject stashing via StashCalculation from unsealed calculations (#7147) [cddeba201]

  • Add support for unstashing (#6826) [6ab3ed1bd]

  • Extend check of stash options when they are specified as empty dict (#6942) [cbbdedb13]

  • RemoteStashCustomData: stashing via custom script (#6777) [b2a6e2132]

CLI#

  • Print worker logfile location in daemon status (#7271) [feb5e80f8]

  • Improve error message clarity for import violations in verdi CLI (#7211) [d1e7778b6]

  • Add orphaned database connection cleanup to verdi process repair (#7207) [1deba397d]

  • Show plugin descriptions in interactive prompt help (#7210) [ff7eec237]

  • Add verdi computer goto command (#7124) [ca78477af]

  • Truncate verdi archive info output node list (#7097) [5b89d1c05]

  • Show first and last node ctime for verdi storage info --detailed (#6829) [f9566a4d4]

Transport#

  • Support 2FA with improved authentication script handling in core.ssh_async transport (#7184) [dc034b146]

  • core.ssh_async: move a technical message from warning to debug (#7024) [8b1943ef3]

  • Set non-interactive default for --safe-interval (#6943) [d0dedc171]

ORM#

  • Support legacy scheduler_state (#7202) [eeb42355b]

  • Add offset to entities.Collection.find (#7125) [d123ac255]

  • Improve typing of QueryBuilder.all() (#6966) [ee87cec5e]

Repository#

  • Make AbstractRepositoryBackend.open() abstract (#7216) [0904796e0]

  • Implement repo methods to extract object size and objects as a zipfile (bytes) (#7126) [d53330c04]

  • Use TextIOWrapper for opening file from repository in text mode (#6847) [761764d81]

Storage#

  • Downgrade SQLite database lock warning to info level (#7121) [1c6c5ac0a]

  • Prior check if migration needed on SqliteZipBackend initialisation (#6963) [7255f01c7]

Fixes#

CLI#

  • verdi status: close broker connection after check (#7269) [b8b77cc7d]

  • Add Discourse and GitHub links to deprecation warnings (#7123) [8aae1b689]

  • Drop dump-related message when relabeling a group (#7033) [239cbf91c]

Engine#

  • asyncio.iscoroutinefunction -> inspect.iscoroutinefunction (#7256) [c4d4a003f]

  • Fix missing return of exit code 0 in inspect_process when not paused (#7247) [2545cb276]

  • Replace deprecated asyncio.get_event_loop() with plumpy.get_or_create_event_loop() (#7206) [f762ecf99]

ORM#

  • TrajectoryData: fix missing space in pbc deprecation warning (#7287) [c39cbfe72]

  • Fix SD cell_angles for zero-length vectors (#6820) [f0a8ce764]

  • TrajectoryData: Fix pbc handling for non-periodic structures (#7079) [08c58ffc0]

  • Fix ResourceWarning from unclosed TextIOWrapper (#7220) [3a061d06e]

Transport#

  • Update error message for SSH transport plugin (#7110) [b2f442276]

  • Fix whoami_async call in sync context (#7083) [16c6ad38e]

  • Unify behavior of core.ssh_async::glob with core.ssh::glob (#6950) [da3e42567]

Storage#

  • Apply SQLA param-limit fix to all IN clauses (#7237) [be0d6b5de]

  • Remove migration output capturing for SqliteZipBackend initialisation (#6964) [016d049ef]

Deprecations#

  • Transport: deprecate chown method (#7178) [be364adfa]

  • Deprecate core.ssh transport plugin in favor of core.ssh_async (#7175) [f9453e738]

  • SshTransport: drop deprecation warnings (#7130) [4a98ec50c]

  • Remove dead code supporting old pymatgen versions (#7066) [17b484fb7]

  • Remove deprecated aiida.common.json module (#7017) [dc702f074]

  • Remove deprecated IPython support (#7013) [2d0a0ed94]

  • Replace aiida.common.assert_never with typing.assert_never (#6856) [494179ddb]

Documentation#

  • Extend XyData documentation (#6736) [f4c205d67]

  • Fix reference to base in put_object_from_file (#7108) [5014979b9]

  • Add how-to page for real-world workflows (#7048) [49af7f0a6]

  • Enhance documentation on process types in AiiDA (#7068) [d50cf17df]

  • Rename section ‘Monitoring processes’ to ‘Inspect processes’ (#7075) [00f082fb3]

  • Remove outdated installation troubleshooting section (#7015) [413b9bd6e]

  • Add unstashing documentation (#6999) [669f249e8]

  • Improve docstring of transport chmod method (#6988) [0eb8fc35f]

  • Full documentation for stashing (#6936) [81dd4df5d]

  • Remove note on unsupported contains and get_creation_statistics (#6930) [a392f5c5c]

Dependencies#

  • Bump the gha-dependencies group across 1 directory with 4 updates (#7253) [e987f7e32]

  • Update uv.lock (#7258) [03df44acc]

  • Bump asyncssh from ~=2.19.0 to ~=2.21.0 (#7240) [93d98ed6e]

  • Bump disk-objectstore to ~=1.5.0 (#7206) [e3217c0fe]

  • Bump the gha-dependencies group across 1 directory with 3 updates (#7194) [16ebbc5d7]

  • Pin mamba version to fix arm64 docker build (#7168) [ca143ca29]

  • uv.lock upgrade (#7077) [7b59996cd]

  • Don’t re-export common.typing, install typing_extensions for all Python versions (#7096) [5c1b2f4cc]

  • Unpin and update bpython (#7031) [f01c70e22]

  • Bump psutil to v7.x (#7012) [c0e3c826d]

  • Support click 8.2 (#6959) [44ef2a17e]

  • Support numpy 2.x (#6899) [1e8cd0b3c]

Devops#

Typing#

  • Add typing to trajectory.py (#7234) [9a6465172]

  • Strict typing for aiida.schedulers module (#7136) [396792b21]

  • Run type checking on engine.processes module (#7114) [a547ea026]

  • Strict typing for aiida.common module (#7085) [289cdd518]

  • Strict typing for aiida.repository module (#7049) [3999e849d]

  • Type check storage.psql_dos.orm.querybuilder.main module (#7065) [baefa764b]

  • Pydantic mypy plugin (#7064) [4455a0bd1]

  • Add typing to aiida.tools.graph module (#7036) [13cb318f1]

  • Add strict typing to aiida.tools.query module (#7041) [34dada2ca]

  • Add typing to aiida/common/{utils,extendeddicts}.py (#6706) [3bae2724b]

  • Run typing on orm/{comments,computers}.py and orm/implementation/storage_backend.py (#6704) [f6cdb91d8]

  • Mark orm.entities.Entity/Comment classes as generic in ruff config (#7019) [ebc9d4ab9]

  • Add more stub packages for type checking 3rd party libs (#7021) [51f0ed718]

  • More typing for cmdline.params module (#7009) [0d5c4d4df]

  • Add typing to cmdline/commands module (#6987) [594cf563c]

  • Typecheck aiida.cmdline.utils module (#6954) [76f776a49]

  • Add typing to aiida.cmdline.params module (#6952) [cfb7fd153]

  • Run type checking on multiply_add.py (#6833) [1229ddd38]

  • Bump mypy to 1.16.1 (#6858) [210684420]

Autogenerated code#

  • Fix script to autogenerate module re-imports and __all__ attributes (#7091) [b3cf7b4c7]

  • Autogenerate aiida.brokers __init__ files, make RabbitmqBroker public (#7127) [f25fc2bea]

  • Autogenerate __all__ in tools module (#7142) [b720897c8]

  • Autogenerate __init__.py re-exports in aiida.transports module (#7137) [4f28b78ba]

CI#

  • Ignore some new DeprecationWarnings in py3.14 (#7256) [4f96f8888]

  • Replace urllib.request.url2pathname with local shim for Python 3.14 (#7240) [1607da454]

  • Pin orm field regression tests to only run for py3.14 (#7240) [ff1871441]

  • Add experimental Python 3.14 support in CI (not yet officially supported; known performance regressions under investigation) (#7240) [e8f39b649]

  • CI: Upload coverage data as GHA artifacts (#7224) [af9d1f600]

  • CI: Change test-install schedule from nightly to weekly (#7179) [bc53c9e31]

  • Refactor bake metadata extraction from shell to JavaScript (#6582) (#7161) [fcf709bc0]

  • Update coveragepy and use sys.monitoring interface (#7042) [1f2cb47cf]

  • CI: Test minimum package versions (#6940) [577d7c92a]

  • Send Slack notification if test-install.yml fails (#6897) [b235b3aab]

Tests#

  • TrajectoryData: add pbc to existing tests and increase coverage (#7234) [e3987509f]

  • test_data: restrict data_plugin fixture to core.* entry points (#7234) [2a3875fa9]

  • Add pytest-instafail; --instafail to addopts (#7203) [605755867]

  • Add --tb=short to default pytest addopts (#7203) [aa6f15e91]

  • Make test_graph_recurse_spot_highlight_classes more robust (#7219) [14f63cd35]

  • Avoid unwanted files in repo during tests (#6977) [00003e3b1]

  • Remove unused code of json-contains tests (#7204) [7536ec298]

  • Fix test isolation in manager fixture by resetting _profile_storage (#7187) [6ee838a4f]

  • Overwrite and sealed unit tests for Group.dump (#7160) [8755758ec]

  • Drop flaky test_json_contains.py benchmarks (#7047) [cd11f08b6]

  • Better coverage for subprocesses (#7070) [ce11608a5]

  • Fix flaky test (#7055) [a843052ac]

  • Tests for orm/utils/node.py + remove dead code (#6983) [ebc49134c]

  • Add more pytest fixture tests (#6882) [6afd3f53d]

  • Cleanup (almost) all warnings in the test suite (#6967) [0eb794e1d]

  • Cleanup verdi tests (#6951) [f99571350]

  • Cleanup warnings in tests (#6911) [cb5348e04]

  • Remove click.Editor.edit_file monkeypatch (#6939) [b4e1c8b83]

Misc#

v2.7.3 - 2026-01-23#

Fixes#

Transport#

  • Improve path escaping in OpenSSH transport for special characters (#7171) [4c89d9624]

  • Fix a critical race condition in TransportQueue (#7144) [b02a233af]

  • Add semaphore control to AsyncSshTransport::exec_command_wait_async to prevent SSH connection overwhelm (#7144) [985d5c809]

Engine#

Devops#

  • Switch from token-based to OIDC trusted publishing [c4e320816]

v2.7.2 - 2025-12-10#

This patch release comes with a number of important bug fixes to the CLI, storage, archive, and transport modules. Most notably, SQLAlchemy operational errors (affecting archive operations and the QB) and various race conditions (affecting mainly the transport plugins) have been resolved. We strongly recommend upgrading from prior 2.7.x versions.

Fixes#

CLI#

  • CLI: Fix verdi code list for codes without computer (#7081) [27b52da2f]

  • CLI: Fix wrongly formatted verdi code show output (#7073) [32742c0e0]

  • CLI: Additional logging during dump operations (#7046) [fc00f5dec]

  • CLI: Drop non-unique -n option for archive import (#7044) [3a7d440e9]

  • CLI: Drop non-unique -p option for dump endpoints (#7043) [019172c2d]

  • CLI: Fix output of verdi node show in dump README files (#6971) [5e4da5b4d]

Storage#

  • Fix RST code snippets in QB smarter_in docstring (#7146) [cc0bb483d]

  • Fix QB IN clause to avoid parameter limits (#6998) [8d562b44e]

  • Fix PSQL OperationalError on archive creation by batching (#6993) [cfbbd687f]

  • Fix OperationalError for add_nodes with PSQL backend (#6991) [9bccdc816]

Archive#

  • Fix UnboundLocalError in ZipfileBackendRepository (#7129) [166d06c25]

  • Fix stuck progress bar on “Add nodes” of archive import (#7118) [4e54cd476]

  • Don’t break on sqlite_zip profile deletion despite original aiida archive file missing (#6929) [274ce6717]

Engine#

  • Adding ./ in front of a portable code binary (#7080) [9256f2fdd]

  • Fix race condition in JobsList (#7061) [e79f0a44c]

Transport#

  • async_ssh: Use async semaphore instead of manual locking (#7018) [ad5cafdb1]

Configuration#

  • Fix: use platform-specific separator for AIIDA_PATH in config directory detection (#6935) [32d515a6b]

Devops#

v2.7.1 - 2025-07-16#

Fixes#

Docs#

v2.7.0 - 2025-06-24#

Asynchronous SSH connection (#6626)#

Previously, when data transfer with a remote computer was active, the responsible transport plugins blocked further program execution until the communication was completed. This long-standing limitation presented a potential opportunity for performance improvements. With the introduction of the new asynchronous SSH transport plugin (core.ssh_async), multiple communications with a remote machine can now happen concurrently. As an added benefit, for the configuration of a Computer with the core.ssh_async transport plugin, it is not necessary anymore to manually provide all SSH connection details during the execution of verdi computer core.ssh_async configure. Instead, one only needs to provide the hostname as it is given in the ~/.ssh/config file, and the transport plugin then uses OpenSSH of the OS to automatically configure the Computer, using the system configuration.

🚀 When core.ssh_async outperforms core.ssh#

core.ssh_async offers significant performance gains in scenarios where the worker is blocked by heavy transfer tasks, such as uploading, downloading, or copying large files.

Example: Submitting two WorkGraphs/WorkChains with the following logic:

  1. WorkGraph 1 – Heavy I/O operations

    • Uploads a 10 MB file

    • Remotely copies a 1 GB file

    • Retrieves a 1 GB file

  2. WorkGraph 2 – Lightweight task

    • Executes a simple shell command: touch file

Measured time until the second WorkGraph is processed (single worker):

  • core.ssh_async: Only 4 seconds! 🚀🚀🚀🚀 A dramatic improvement!

  • core.ssh: 108 seconds (the second task waits for the first to finish)

⚖️ When core.ssh_async and core.ssh perform similarly#

For mixed workloads involving numerous uploads and downloads—a common real-world use case—the performance gains depend on the specific conditions.

Large file Transfers (~1 GB):#

core.ssh_async typically outperforms due to concurrent upload and download streams. In favorable network conditions, this can nearly double the effective bandwidth.

Example: On a network with a baseline of 11.8 MB/s, the asynchronous mode approached nearly twice that speed under light load (see graph in PR #6626).

Test case:
Two WorkGraphs: one uploads 1 GB, the other retrieves 1 GB using RemoteData.

  • core.ssh_async: 120 seconds

  • core.ssh: 204 seconds

Small file transfers (many small files):#

Here, the overhead of managing asynchronous operations can outweigh the benefits.

Test case:
25 WorkGraphs, each transferring several ~1 MB files.

  • core.ssh_async: 105 seconds

  • core.ssh: 65 seconds

To conclude, the choice of which transport plugin is the best bet for your use case depends on your specific application: use core.ssh_async for workloads involving large file transfers or when you need to prevent I/O operations from blocking other tasks, but stick with core.ssh for scenarios dominated by many small file transfers where the asynchronous overhead may reduce performance.

Extended dumping support for profiles and groups (#6723)#

In version v2.6.0, AiiDA introduced the ability to dump processes from the database into a human-readable, structured folder format. Building on this feature, support has now been extended to allow dumping of entire groups and profiles, enabling users to retrieve AiiDA data more easily. This enhancement is part of our broader roadmap to improve AiiDA’s usability—especially for new users—who may find it challenging to construct the appropriate queries to extract data from the database manually. The functionality is accessible via the verdi CLI:

verdi profile dump --all          # This dumps the whole current profile
verdi profile dump --groups <PK>  # This dumps one selected group as part of the profile dumping operation
verdi group dump <PK>             # This dumps only the selected group, disregarding other profile data

Since dumping an entire profile can be a resource- and I/O-intensive operation (for large profiles), significant effort has been made to provide flexible options for fine-tuning which nodes are included in the dump. To avoid initiating dumping operations on large profiles, if no filters (e.g., groups, codes, computers, node mtime, etc.) are set, by default, no data is being dumped. If all data of a profile should be dumped, this must be actively requested using the --all option, or one must explicitly select the subset of data to be included in the dump via the provided filter options. Below is a snippet from the command’s help output:

Usage: verdi profile dump [OPTIONS] [--]

  Dump all data in an AiiDA profiles storage to disk.

Options:
  -p, --path PATH                 Base path for dump operations that write to
                                  disk.
  -n, --dry-run                   Perform a dry run.
  -o, --overwrite                 Overwrite file/directory when writing to
                                  disk.
  -a, --all                       Include all entries, disregarding all other
                                  filter options and flags.
  -X, --codes CODE...             One or multiple codes identified by their
                                  ID, UUID or label.
  -Y, --computers COMPUTER...     One or multiple computers identified by
                                  their ID, UUID or label.
  -G, --groups GROUP...           One or multiple groups identified by their
                                  ID, UUID or label.
  -u, --user USER                 Email address of the user.
  -p, --past-days PAST_DAYS       Only include entries created in the last
                                  PAST_DAYS number of days.
  --start-date TEXT               Start date for node mtime range selection
                                  for node collection dumping.
  --end-date TEXT                 End date for node mtime range selection for
                                  node collection dumping.
  --filter-by-last-dump-time / --no-filter-by-last-dump-time
                                  Only select nodes whose mtime is after the
                                  last dump time.  [default: filter-by-last-
                                  dump-time]
  --only-top-level-calcs / --no-only-top-level-calcs
                                  Dump calculations in their own dedicated
                                  directories, not just as part of the dumped
                                  workflow.  [default: only-top-level-calcs]
  --only-top-level-workflows / --no-only-top-level-workflows
                                  If a top-level workflow calls sub-workflows,
                                  create a designated directory only for the
                                  top-level workflow.  [default: only-top-
                                  level-workflows]
  --delete-missing / --no-delete-missing
                                  If a previously dumped group or node is
                                  deleted from the DB, delete the
                                  corresponding dump directory.  [default:
                                  delete-missing]
  --symlink-calcs / --no-symlink-calcs
                                  Symlink workflow sub-calculations to their
                                  own dedicated directories.  [default: no-
                                  symlink-calcs]
  --organize-by-groups / --no-organize-by-groups
                                  If the collection of nodes to be dumped is
                                  organized in groups, reproduce its
                                  hierarchy.  [default: organize-by-groups]
  --also-ungrouped / --no-also-ungrouped
                                  Dump also data of nodes that are not part of
                                  any group.  [default: no-also-ungrouped]
  --relabel-groups / --no-relabel-groups
                                  Update directories and log entries for the
                                  dumping if groups have been relabeled since
                                  the last dump.  [default: relabel-groups]
  --include-inputs / --exclude-inputs
                                  Include linked input nodes of
                                  `CalculationNode`(s).  [default: include-
                                  inputs]
  --include-outputs / --exclude-outputs
                                  Include linked output nodes of
                                  `CalculationNode`(s).  [default: exclude-
                                  outputs]
  --include-attributes / --exclude-attributes
                                  Include attributes in the
                                  `aiida_node_metadata.yaml` written for every
                                  `ProcessNode`.  [default: include-
                                  attributes]
  --include-extras / --exclude-extras
                                  Include extras in the
                                  `aiida_node_metadata.yaml` written for every
                                  `ProcessNode`.  [default: exclude-extras]
  -f, --flat                      Dump files in a flat directory for every
                                  step of a workflow.
  --dump-unsealed / --no-dump-unsealed
                                  Also allow the dumping of unsealed process
                                  nodes.  [default: no-dump-unsealed]
  -v, --verbosity [notset|debug|info|report|warning|error|critical]
                                  Set the verbosity of the output.
  -h, --help                      Show this message and exit.

Another key feature is the incremental nature of the command, which ensures that the dumping process synchronizes the output folder with the internal state of AiiDA’s DB by gradually adding or removing files on successive executions of the command. This allows for efficient updates without having to overwrite everything, and is in contrast to AiiDA archive creation, which is a one-shot process. The behavior can further be adjusted using:

  • --dry-run (-n): to simulate the dump without writing any files.

  • --overwrite (-o): to fully overwrite the target directory if it already exists.

Finally, the command provides various options to customize the output folder structure, for instance, to reflect the group hierarchy of AiiDA’s internal DB state, symlink duplicate calculations (e.g., which are contained in multiple groups), create dedicated directories for sub-workflows and calculations of top-level workflows, and more.

These enhancements aim to make data export from AiiDA more robust, customizable, and user-friendly.

Stashing (#6746, #6772)#

With this feature, you can bundle your data to a (compressed) tar archive during stashing by specifying one of the stash_mode options "tar", "tar.bz2", "tar.gz", or "tar.xz". When specifying the stashing operation during the setup of your calculation, compression can be configured as follows:

from aiida.plugins import CalculationFactory
from aiida.engine import run
from aiida.common import StashMode
from aiida.orm import load_computer

inputs = {
    ...,
    'metadata': {
        'computer': load_computer(label="localhost"),
        'options': {
            'resources': {'num_machines': 1},
            'stash': {
                'stash_mode':  StashMode.COMPRESS_TARGZ.value,
                'target_base': '/scratch/',
                'source_list': ['heavy_data.xyz'],  # ['*'] to stash everything
            },
        },
    },
}
# If you use a builder, use
# builder.metadata = {'options': {...}, ...}

run(MyCalculation, **inputs)

In addition, it was historically only possible to enable stashing when it was instructed before running a generic CalcJob. This means that the instruction had to be “attached” to the original CalcJob before its execcution. However, if a user would realize they need to stash something only after running the calculation, this would not be possible. With v2.7.0, we introduce the new StashCalculation CalcJob which is able to perform a stashing operation after a calculation has finished—provenance included! The usage is very similar, and for consistency and user-friendliness, we keep the instructions as part of the metadata. The only main input is the remote_folder output node (an instance of RemoteData) of the calculation source node to be stashed, for example:

from aiida.plugins import CalculationFactory
from aiida.engine import run
from aiida.common import StashMode
from aiida.orm import load_node

StashCalculation = CalculationFactory('core.stash')

calcjob_node = load_node(<CALCJOB_PK>)
inputs = {
    'metadata': {
        'computer': calcjob_node.computer,
        'options': {
            'resources': {'num_machines': 1},
            'stash': {
                'stash_mode':  StashMode.COPY.value,
                'target_base': '/scratch/',
                'source_list': ['heavy_data.xyz'],
            },
        },
    },
    'source_node': calcjob_node.outputs.remote_folder,
}

result = run(StashCalculation, **inputs)

Forcefully killing processes (#6793)#

Prior to version v2.7.0, the verdi process kill command could hang if a connection to the remote computer could not be established. A new --force option has been introduced to terminate a process without waiting for a response from the remote machine.
Note: Using --force may result in orphaned jobs on the remote system if the remote job cancellation fails.

verdi process kill --force <PROCESS_ID>

We also now cancel the old killing action if it is resend by the user. This allows the user to adapt the parameters for the exponential backoff mechanism (EBM) applied by AiiDA in the verdi config and then resend the kill command with the new parameters.

verdi process kill --timeout 5 <PROCESS_ID>
verdi config set transport.task_maximum_attempts 1
verdi config set transport.task_retry_initial_interval 5
verdi daemon restart
verdi process kill <PROCESS_ID>

Furthermore, the timeout and wait options were not behaving correctly, so they are now fixed and both merged into the single timeout option. By passing --timeout 0 it replicates the --no-wait functionality, meaning the command does not block until the action has finished, and by passing --timeout inf (default option, replicating --wait without a timeout), the command blocks until a response. For more information see issue #6524.

Serialization of ORM nodes (#6723)#

AiiDA’s Python API provides an object relational mapper (ORM) that abstracts the various entities that can be stored inside the provenance graph (via the SQL database) and the relationships between them. In most use cases, users use this ORM directly in Python to construct new instances of entities and retrieve existing ones, in order to get access to their data and manipulate it. A shortcoming of the current ORM is that it is not possible to programmatically introspect the schema of each entity: that is to say, what data each entity stores. This makes it difficult for external applications to provide interfaces to create and or retrieve entity instances. It also makes it difficult to take the data outside of the Python environment since the data would have to be serialized. However, without a well-defined schema, doing this without an ad-hoc solution is practically impossible.

With the implementation of a pydantic Model for each Entity we now allow external applications to programmatically determine the schema of all AiiDA ORM entities and automatically (de)serialize entity instances to and from other data formats, e.g., JSON. An example how this is done for an AiiDA integer node:

node = Int(5) # Can be any ORM node
serialized_node = node.serialize()
print(serialized_node)
# Out: {'pk': None, 'uuid': '485c2ec8-441d-484d-b7d9-374a3cdd98ae', 'node_type': 'data.core.int.Int.', 'process_type': None, 'repository_metadata': {}, 'ctime': datetime.datetime(2025, 5, 2, 10, 20, 41, 275443, tzinfo=datetime.timezone(datetime.timedelta(seconds=7200), 'CEST')), 'mtime': None, 'label': '', 'description': '', 'attributes': {'value': 5}, 'extras': {}, 'computer': None, 'user': 1, 'repository_content': {}, 'source': None, 'value': 5}
uuid: 77e9c19a-5ecb-40cf-8238-ea5c55fbb83f (unstored) value: 5
node_deserialized = Int.from_serialized(**serialized_node)
print(node_deserialized)
# Out: uuid: 77e9c19a-5ecb-40cf-8238-ea5c55fbb83f (unstored) value: 5

For an extensive overview of the implications see AEP 010.

Miscellaneous#

  • aiida-core is now compatible with Python 3.13 #6600

  • OpenSSH backend for transport core.ssh_async for multiplexing #6795

  • Improved Windows support #6715

  • RemoteData extended by member function get_size_on_disk #6584

  • SinglefileData extended by constructor from_bytes #6653

  • Allow zero memory specification for SLURM #6605

  • Add filters to verdi group delete #6556

  • verdi storage maintain shows a progress bar #6562

  • New transport endpoints compress & extract #6743

  • Implementation of missing SQLite endpoints (en route to full feature parity between PostgreSQL and SQLite):

Full list of changes for users#

Features#

  • CLI: Accept mulitple node identifiers in verdi node graph generate (#6443) [6d2edc919]

  • CLI: Add default for output_file in computer and code export commands (#6486) [9355a9878]

  • CLI: Validate storage in verdi storage version (#6551) [ad1a431f3]

  • CLI: Add filters to verdi group delete. (#6556) [72a6b183b]

  • CLI: verdi storage maintain: show a progress bar (#6562) [c7c289d38]

  • implement has_key filter for SQLite backend [779cc29d8]

  • ORM: Add from_bytes classmethod to orm.SinglefileData (#6653) [0f0b88a39]

  • ORM: Add get_size_on_disk method to RemoteData (#6584) [02cbe0ceb]

  • QB: Re-introduction of contains Filter Operator for SQLite (#6619) [aa0aa262a]

  • Transport: AsyncTransport plugin (#6626) [eba6954bf]

  • Add support for running direct jobs on bash environment on Windows [ce9dcf421]

  • Transport: feat compress & extract methods (#6743) [f4c55f5f7]

  • CLI: add option --clean-workdir to verdi node delete (#6756) [c53592850]

  • RemoteStashCompressedData new data class, and it’s deployment to execmanager.py to support compressed file formats while stashing. [ae49af6c3]

  • [QueryBuilder] Implement get_creation_statistics for SQLite backend (#6763) [83454c713]

  • Support for Python 3.13 (#6600) [eb34b0606]

  • StashCalculation: a new CalcJob plugin (#6772) [bc253236d]

  • ORM: Use pydantic to specify a schema for each ORM entity (#6255) [958bfd05c]

  • Add force-kill option when killing a process (#6793) [b6d0fe50d]

  • Align the yes/no prompt in verdi computer delete with other prompts [71fc14f3c2a501ff8d704d20df76a297edc8e8bc]

  • Profile data dumping (#6723) [88df72a46073359bd4bdaa03cb5f4c5a92aa67af]

  • Transport: Add OpenSSH as backend option to AsyncSshTransport (#6795) [c7fdf1cfaf50b555698ef039d55b7f295a71644a]

Fixes#

Documentation#

Full list of changes for developers#

Source code#

Tests#

Devops#

  • Devops: Update pre-commit dependencies (#6504) [14bb05f4b]

  • Docker: Fix release tag in publish workflow (#6520) [c740b99f2]

  • Devops: Mark test_leak_ssh_calcjob as nightly (#6521) [a5da4eda1]

  • Devops: Add type hints to aiida.orm.utils.remote (#6503) [2bdcb7f00]

  • Docker: Make warning test insensitive to deprecation warnings (#6541) [f1be224c4]

  • CI: Update ignore comment as the way that presumably updated mypy expects (#6566) [655da5acc]

  • Update the issue template with the Discourse group (#6588) [a7d8dd04e]

  • Fix broken troubleshooting link in bug report issue template (#6589) [f57591655]

  • Devops: Add tox environment for pytest with presto mark [e2699118e]

  • CLI: Dump only sealed process nodes (#6591) [70572380b]

  • CLI: Check user execute in verdi code test for InstalledCode (#6597) [8350df0cb]

  • Devops: Bump peter-evans/create-pull-request (#6576) [dd866ce81]

  • Bump mypy version to ~=1.13.0 (#6630) [c93fb4f75]

  • CI: remove ci-style.yml (#6638) [451375b7c]

  • Pre-commit: exclude mypy for all tests (#6639) [846c11f8c]

  • Add helper script for creating patch releases from a list of commits (#6602) [ec8a055a5]

  • CLI: Handle None process states in build_call_graph (#6590) [f74adb94c]

  • Bump ruff version (#6614) [c915a9734]

  • DevOp: Using xdist to run pytest in parallel (#6620) [090dc1c73]

  • Bump codecov/codecov-action from 4 to 5 in the gha-dependencies group (#6648) [835d13b73]

  • Amend type call error after using AiiDAConfigDir (#6646) [dbdc36c63]

  • CI: Turn off verbose pytest output (#6633) [41a0fd92b]

  • Bump ruff to v0.8.0 (#6634) [333992be5]

  • CI: Utilize uv lockfile for reproducible test environments (#6640) [04cc34488]

  • CI: Test with RabbitMQ 4.0.x (#6649) [a1872b1e7]

  • CI: quick fix on failed benchmark CI using uv run pytest (#6652) [37e5431e6]

  • Typing-extensions as dependency for py3.9 (#6664) [c532b34a1]

  • Update uv.lock + CI tweaks + fix pymatgen issue (#6676) [10930266f]

  • Devops: Update pre-commit dependencies (#6683) [0eb77b8ae]

  • CI: Add matrix testing for both SQLite and PostgreSQL database backends [a3ccec96d]

  • Fix redundant “tests/” in test-install.yml (#6695) [5e8bbe1ae]

  • CI: Drop workaround to pass pymatgen related failed tests (#6694) [d17c2931a]

  • Devops: Add explicit sphinx.configuration key to RTD conf (#6700) [8440416a5]

  • Use uv lockfile in readthedocs build (#6685) [15b5caf23]

  • Run mypy on src/aiida/orm/nodes/caching.py (#6703) [199a0276c]

  • CI: Set runners to ubuntu-24.04 (#6696) [b43261143]

  • CI: fix failed test_containerized.py integration test for containerized code (#6707) [738705611]

  • Nightly tests adjust run test_memory_leak in test move high_link to nightly (#6701) [b3569508c]

  • Set minimum uv version in pyproject.toml (#6714) [c88fc05ba]

  • Use uv-pre-commit to validate lockfile (#6699) [208d6a967]

  • Use Github arm runners for docker arm build tests (#6717) [f43a51010]

  • CI: Ignore mamba Warning Message on stderr to Prevent JSON Parsing Errors (#6748) [8c5c709be]

  • Fix docker arm build (#6759) [c4dfadabf]

  • CI: RTD tail errors and warnings (#6776) [bb5f93daa]

  • Bump the gha-dependencies group with 2 updates (#6778) [48bd2acbc]

  • Revert “Bump the gha-dependencies group with 2 updates (#6778)” (#6838) [61be15d54]

  • Add Python 3.13 to tests job in test-install workflow and for verdi and presto jobs in ci-code workflow (#6843) [6cb2c712d]

  • Add back pre-commit job for mypy type-checking (#6827) [6eb17a7d0]

  • RTD: add a developer comment in .readthedocs.yml (#6864) [5807a390927a5892cd349bf8e3f3da5ef233eca5]

  • Devops: Skip upload step in benchmarks if not pushed to main (#6895) [6b5578602b0f00b22ac457c070c764168ac8cbd4]

  • Devops: Upgrade uv>=v0.7.6 and astral-sh/setup-uv@v6 (#6887) [5fbfe12f2a1805c0c69ee9c77f1a0d29106b1aad]

  • Upgrade uv lockfile (#6886) [edb8c7e88fde03510e05fb9cc48cc58aa7112dc5]

v2.6.4 - 2025-05-22#

Fixes#

  • Dependencies: Pin click to ~=8.1.0 (#6888) [e1d55fa]

v2.6.3 - 2024-11-06#

Fixes#

  • CLI: Fix exception for verdi plugin list (#6560) [c3b10b7]

  • DirectScheduler: Ensure killing child processes (#6572) [fddffca]

  • Engine: Fix state change broadcast before process node is updated (#6580) [867353c]

Devops#

  • Docker: Replace sleep with s6-notifyoncheck (#6475) [9579378b]

  • Fix failed docker CI using more reasoning grep regex to parse python version (#6581) [332a4a91]

  • DevOps: Fix json query in reading the docker names to filter out fields not starting with aiida (#6573) [e1467edc]

v2.6.2 - 2024-08-07#

Fixes#

  • LocalTransport: Fix typo for ignore_nonexisting in put (#6471) [ecda558d0]

  • CLI: verdi computer test report correct failed tests (#6536) [9c3f2bb58]

  • CLI: Fix verdi storage migrate for profile without broker (#6550) [389fc487d]

  • CLI: Fix bug verdi presto when tab-completing without config (#6535) [efcf75e40]

  • Engine: Change signature of set_process_state_change_timestamp [923fd9f6e]

  • Engine: Ensure node is sealed when process excepts (#6549) [e3ed9a2f3]

  • Engine: Fix bug in upload calculation for PortableCode with SSH (#6519) [740ae2040]

  • Engine: Ignore failing process state change for core.sqlite_dos [fb4f9815f]

Dependencies#

  • Pin requirement to minor version sphinx~=7.2.0 (#6527) [25cb73188]

Documentation#

  • Add PluggableSchemaValidator to nitpick exceptions (#6515) [0ce3c0025]

  • Add robots.txt to only allow indexing of latest and stable (#6517) [a492e3492]

  • Add succint overview of limitations of no-services profile [d7ca5657b]

  • Correct signature of get_daemon_client example snippet (#6554) [92d391658]

  • Fix typo in pytest plugins codeblock (#6513) [eb23688fe]

  • Move limitations warning to top of quick install [493e529a7]

  • Remove note in installation guide regarding Python requirement [9aa2044e4]

  • Update redirects.txt for installation pages (#6509) [508a9fb2a]

Devops#

  • Fix pymatgen import causing mypy to fail (#6540) [813374fe1]

v2.6.1 - 2024-07-01#

Fixes#

  • Fixtures: Make pgtest truly an optional dependency [9fe8fd2e0]

v2.6.0 - 2024-07-01#

This minor release comes with a number of features that are focused on user friendliness and ease-of-use of the CLI and the API. The caching mechanism has received a number of improvements guaranteeing even greater savings of computational time. For existing calculations to be valid cache sources in the new version, their hash has to be regenerated (see Improvements and changes to caching for details).

Making RabbitMQ optional#

The RabbitMQ message broker service is now optional for running AiiDA. The requirement was added in AiiDA v1.0 when the engine was completely overhauled. Although it significantly improved the scaling and responsiveness, it also made it more difficult to start using AiiDA. As of v2.6, profiles can now be configured without RabbitMQ, at the cost that the daemon can not be used and all processes have to be run locally.

Simplifying profile setup#

With the removal of RabbitMQ as a hard requirement, combined with storage plugins that replace PostgreSQL with the serverless SQLite that were introduced in v2.5, it is now possible to setup a profile that requires no services. A new command is introduced, verdi presto, that automatically creates a profile with sensible defaults. This now in principle makes it possible to run just the two following commands on any operating system:

pip install aiida-core
verdi presto

and get a working AiiDA installation that is ready to go. As a bonus, it also configures the localhost as a Computer. See the documentation for more details.

Improved test fixtures without services#

Up till now, running tests would always require a fully functional profile, which meant that PostgreSQL and RabbitMQ had to be available. As described in the section above, it is now possible to set up a profile without these services. This new feature is leveraged to provide a set of pytest fixtures that provide a test profile that can be used essentially on any system that just has AiiDA installed. To start writing tests, simply create a conftest.py and import the fixtures with:

pytest_plugins = 'aiida.tools.pytest_fixtures'

The new fixtures include the aiida_profile fixture which is session-scoped and automatically loaded. The fixture creates a temporary test profile at the start of the test session and automatically deletes it when the session ends. For more information and an overview of all available fixtures, please refer to the documentation on pytest fixtures.

Improvements and changes to caching#

A number of fixes and changes to the caching mechanism were introduced (see the changes subsection of the full list of changes for a more detailed overview). For existing calculations to be valid cache sources in the new version, their hash has to be regenerated by running verdi node rehash. Note that this can take a while for large databases.

Since its introduction, the cache would essentially be reset each time AiiDA or any of the plugin packages would be updated, since the version of these packages were included in the calculation of the node hashes. This was originally done out of precaution to err on the safe-side and limit the possibility of false-positives in cache hits. However, this strategy has turned out to be unnecessarily cautious and severely limited the effectiveness of caching.

The package version information is no longer included in the hash and therefore no longer impacts the caching. This change does now make it possible for false positives if the implementation of a CalcJob or Parser plugin changes signficantly. Therefore, a mechanism is introduced to give control to these plugins to effectively reset the cache of existing nodes. Please refer to the documentation on controlling caching for more details.

Programmatic syntax for query builder filters and projections#

In the QueryBuilder, fields to filter on or project always had to be provided with strings:

QueryBuilder().append(Node, filters={'label': 'some-label'}, project=['id', 'ctime'])

and it is not always trivial to know what fields exist that can be filtered on and can be projected. In addition, there was a discrepancy for some fields, most notably the pk property, which had to be converted to id in the query builder syntax.

These limitations have been solved as each class in AiiDA’s ORM now defines the fields property, which allows to discover these fields programmatically. The example above would convert to:

QueryBuilder().append(Node, filters={Node.fields.label: 'some-label'}, project=[Node.fields.pk, Node.fields.ctime])

The fields property provides tab-completion allowing easy discovery of available fields for an ORM class in IDEs and interactive shells. The fields also allow to express logical conditions programmatically and more. For more details, please refer to the documentation on programmatic field syntax.

Data plugins can also define custom fields, adding on top of the fields inherited from their base class(es). The documentation on data plugin fields provides more information, but the API is currently in beta and guaranteed to be changed in an upcoming version. It is therefore recommended for plugin developers to hold off making use of this new API.

Automated profile storage backups#

A generic mechanism has been implemented to allow easily backing up the data of a profile. The command verdi storage backup automatically maintains a directory structure of previous backups allowing efficient incremental backups. Note that the exact details of the backup mechanism is dependent on the storage plugin that is used by the profile and not all storage plugins necessarily implement it. For now the storage plugins core.psql_dos, and core.sqlite_dos implement the functionality. For more information, please refer to the documentation. Please refer to this section of the documentation for instructions to restore from a backup.

Full list of changes#

Features#

  • CalcJob: Allow to define order of copying of input files [6898ff4d8]

  • SqliteDosStorage: Implement the backup functionality [18e447c77]

  • SshTransport: Return FileNotFoundError if destination parent does not exist [d86bb38bf]

  • Add improved more configurable versions of pytest fixtures [e3a60460e]

  • Add the orm.Entity.fields interface for QueryBuilder [4b9abe2bd]

  • CLI: verdi computer test make unexpected output check optional [589a3b2c0]

  • CLI: verdi node graph generate root nodes as arguments [06f8f4cfb]

  • CLI: Add --most-recent-node option to verdi process watch [72692fa5c]

  • CLI: Add --sort/--no-sort to verdi code export [80c606890]

  • CLI: Add verdi process dump and the ProcessDumper [6291accf0]

  • CLI: Add RabbitMQ options to verdi profile setup [f553f805e]

  • CLI: Add the -M/--most-recent-node option [5aae874aa]

  • CLI: Add the verdi computer export command [9e3ebf6ea]

  • CLI: Add the verdi node list command [cf091e80f]

  • CLI: Add the verdi presto command [6b6e1520f]

  • CLI: Add the verdi profile configure-rabbitmq command [202a3ece9]

  • CLI: Allow verdi computer delete to delete associated nodes [348777571]

  • CLI: Allow multiple root nodes in verdi node graph generate [f16c432af]

  • Engine: Allow CalcJob monitors to return outputs [b7e59a0db]

  • Make postgres_cluster and config_psql_dos fixtures configurable [35d7ca63b]

  • Process: Add the metadata.disable_cache input [4626b11f8]

  • Storage: Add backup mechanism to the interface [bf79f23ee]

  • Transports: fix overwrite behaviour for puttree/gettree [a55451703]

Performance#

  • CLI: Speed up tab-completion by lazily importing Config [9524cda0b]

  • Improve import time of aiida.orm and aiida.storage [fb9b6cc3b]

  • ORM: Cache the logger adapter for ProcessNode [1d104d06b]

Changes#

  • Caching: NodeCaching._get_objects_to_hash return type to dict [c9c7c4bd8]

  • Caching: Add CACHE_VERSION attribute to CalcJob and Parser [39d0f312d]

  • Caching: Include the node’s class in objects to hash [68ce11161]

  • Caching: Make NodeCaching._get_object_to_hash public [e33000402]

  • Caching: Remove core and plugin information from hash calculation [4c60bbef8]

  • Caching: Rename get_hash to compute_hash [b544f7cf9]

  • CLI: Always do hard reset in verdi daemon restart [8ac642410]

  • CLI: Change --profile to -p/--profile-name for verdi profile setup [8ea203cd9]

  • CLI: Let -v/--verbosity only affect aiida and verdi loggers [487c6bf04]

  • Engine: Set the to_aiida_type as default inport port serializer [2fa7a5305]

  • QueryBuilder: Remove implementation for has_key in SQLite storage [24cfbe27e]

Fixes#

  • BandsData: Use f-strings in _prepare_gnuplot [dba117437]

  • BaseRestartWorkChain: Fix handler overrides used only first iteration [65786a6bd]

  • SlurmScheduler: Make detailed job info fields dynamic [4f9774a68]

  • SqliteDosStorage: Fix exception when importing archive [af0c260bb]

  • StructureData: Fix the pbc constraints of get_pymatgen_structure [adcce4bcd]

  • Archive: Automatically create nested output directories [212f6163b]

  • Archive: Respect filter_size in query for existing nodes [ef60b66aa]

  • CLI: Ensure deprecation warnings are printed before any prompts [deb293d0e]

  • CLI: Fix verdi archive create --dry-run for empty file repository [cc96c9d04]

  • CLI: Fix verdi plugin list incorrectly not displaying description [e952d7717]

  • CLI: Fix verdi process [show|report|status|watch|call-root] no output [a56a1389d]

  • CLI: Fix verdi process list if no available workers [b44afcb3c]

  • CLI: Fix verdi quicksetup when profiles exist where storage is not core.psql_dos [6cb91c181]

  • CLI: Fix dry-run resulting in critical error in verdi archive import [36991c6c8]

  • CLI: Fix logging not showing in verdi daemon worker [9bd8585bd]

  • CLI: Fix the ctx.obj.profile attribute not being initialized [8a286f26e]

  • CLI: Hide misleading message for verdi archive create --test-run [7e42d7aa7]

  • CLI: Improve error message of PathOrUrl and FileOrUrl [ffc6e4f70]

  • CLI: Only configure logging in set_log_level callback once [66a2dcedd]

  • CLI: Unify help of verdi process commands [d91e0a58d]

  • Config: Set existing user as default for read-only storages [e66592509]

  • Config: Use UUID in Manager.load_profile to identify profile [b01038bf1]

  • Daemon: Log the worker’s path and Python interpreter [ae2094169]

  • Docker: Start and stop daemon only when a profile exists [0a5b20023]

  • Engine: Add positional inputs for Process.submit [d1131fe94]

  • Engine: Catch NotImplementedErrorin get_process_state_change_timestamp [04926fe20]

  • Engine: Fix paused work chains not showing it in process status [40b22d593]

  • Fix passwords containing @ not being accepted for Postgres databases [d14c14db2]

  • ORM: Correct field type of InstalledCode and PortableCode models [0079cc1e4]

  • ORM: Fix ProcessNode.get_builder_restart [0dee9d8ef]

  • ORM: Fix deprecation warning being shown for new code types [a9155713b]

  • Runner: Close event loop in Runner.close() [53cc45837]

Deprecations#

  • CLI: Deprecate verdi profile setdefault and rename to verdi profile set-default [ab48a4f62]

  • CLI: Deprecate accepting 0 for default_mpiprocs_per_machine [acec0c190]

  • CLI: Deprecate the deprecated_command decorator [4c11c0616]

  • CLI: Remove the deprecated verdi database command [3dbde9e31]

  • ORM: Undo deprecation of Code.get_description [1b13014b1]

Dependencies#

Refactoring#

  • Abstract message broker functionality [69389e038]

  • Config: Refactor get_configuration_directory_from_envvar [65739f524]

  • Config: Refactor the create_profile function and method [905e93444]

  • Engine: Refactor handling of remote_folder and retrieved outputs [28adacaf8]

  • ORM: Switch to pydantic for code schema definition [06189d528]

  • Replace deprecated IOError with OSError [7f9129fd1]

  • Storage: Move profile locking to the abstract base class [ea5f51bcb]

Documentation#

Devops#

  • Add check-jsonschema pre-commit hook for GHA workflows [14c5bb0f7]

  • Add Dependabot config for maintaining GH actions [0812f4b9e]

  • Add docker image aiida-core-dev for development [6d0984109]

  • Add Python 3.12 tox environment [6b0d43960]

  • Add the slurm service to nightly workflow [5460a0414]

  • Add typing to aiida.common.hashing [ba21ba1d4]

  • Add workflow to build Docker images on PRs from forks [23d2aa5ee]

  • Address internal deprecation warnings [ceed7d55d]

  • Allow unit test suite to be ran against SQLite [0dc8bbcb2]

  • Bump the gha-dependencies group with 4 updates [ccb56286c]

  • Dependencies: Update the requirements files [61ae1a55b]

  • Disable code coverage in test-install.yml [4cecda517]

  • Do not pin the mamba version [82bba1307]

  • Fix Docker build not defining REGISTRY [e7953fd4d]

  • Fix publishing to DockerHub using incorrect secret name [9c9ff7986]

  • Fix Slack notification for nightly tests [082589f45]

  • Fix the test-install.yml workflow [22ea06362]

  • Fix the Docker builds [3404c0192]

  • Increase timeout for the test-install workflow [e36a3f11f]

  • Move RabbitMQ CI to nightly and update versions [b47a56698]

  • Refactor the GHA Docker build [e47932ee9]

  • Remove verdi tui from CLI reference documentation [1b4a19a44]

  • Run Docker workflow only for pushes to origin [b1a714155]

  • Tests: Convert hierarchy functions into fixtures [a02abc470]

  • Tests: extend node_and_calc_info fixture to core.ssh [9cf28f208]

  • Tests: Remove test classes for transport plugins [b77e51f8c]

  • Tests: Unskip test in tests/cmdline/commands/test_archive_import.py [7b7958c7a]

  • Update codecov action [fc2a84d9b]

  • Update deprecated whitelist_externals option in tox config [8feef5189]

  • Update pre-commit hooks [3dda84ff3]

  • Update pre-commit requirement ruff==0.3.5 [acd54543d]

  • Update requirements mypy and pre-commit[04b3260a0]

  • Update requirements to address deprecation warnings [566f681f7]

  • Use uv to install package in CI and CD [73a734ae3]

  • Use recursive dependencies for pre-commit extra [6564e78dd]

v2.5.1 - 2024-01-31#

This is a patch release with a few bug fixes, but mostly devops changes related to the package structure.

Fixes#

  • CLI: Fix verdi process repair not actually repairing [784ad6488]

  • Docker: Allow default profile parameters to be configured through env variables [06ea130df]

Dependencies#

  • Dependencies: Fix incompatibility with spglib>=2.3 [fa8b9275e]

Devops#

  • Devops: Move the source directory into src/ [53748d4de]

  • Devops: Remove post release action for uploading pot to transifex [9feda35eb]

  • Pre-commit: Add ruff as the new linter and formatter [64c5e6a82]

  • Pre-commit: Update a number of pre-commit hooks [a4ced7a67]

  • Pre-commit: Add YAML and TOML formatters [c27aa33f3]

  • Update pre-commit CI configuration [cb95f0c4c]

  • Update pre-commit dependencies [8dfab0e09]

  • Dependencies: Pin mypy to minor version mypy~=1.7.1 [d65fa3d2d]

Documentation#

  • Streamline and fix typos in docs/topics/processes/usage.rst [45ba27732]

  • Update process function section on file deduplication [f35d7ae98]

  • Correct a typo in docs/source/topics/data_types.rst [6ee278ceb]

  • Fix the ADES paper citation [80117f8f7]

v2.5.0 - 2023-12-20#

This minor release comes with a number of features that are focused on user friendliness of the CLI and the API. It also reduces the import time of modules, which makes the CLI faster to load and so tab-completion should be snappier. The release adds support for Python 3.12 and a great number of bugs are fixed.

Create profiles without a database server#

A new storage backend plugin has been added that uses SQLite instead of PostgreSQL. This makes it a lot easier to setup across all platforms. A new profile using this storage backend can be created in a single command:

verdi profile setup core.sqlite_dos -n --profile <PROFILE_NAME> --email <EMAIL>

Although easier to setup compared to the default storage backend that uses PostgreSQL, it is less performant. This makes this storage ideally suited for use-cases that want to test or demonstrate AiiDA, or to just play around a bit. The storage is compatible with most of AiiDA’s functionality, except for automated database migrations and some very specific QueryBuilder functionality. Therefore, for production databases, the default core.psql_dos storage entry point remains the recommended storage.

It is now also possible to create a profile using an export archive:

verdi profile setup core.sqlite_dos -n --profile <PROFILE_NAME> --filepath <ARCHIVE>

where <ARCHIVE> should point to an export archive on disk. You can now use this profile like any other profile to inspect the data of the export archive. Note that this profile is read-only, so you will not be able to use it to mutate existing data or add new data to the profile. See the documentation for more details and a more in-depth example.

Finally, the original storage plugin core.psql_dos, which uses PostgreSQL for the database is also accessible through verdi profile setup core.psql_dos. Essentially this is the same as the verdi setup command, which is kept for now for backwards compatibility.

See the documentation on storage plugins for more details on the differences between these storage plugins and when to use which.

The verdi profile delete command can now also be used to delete a profile for any of these storage plugins. You will be prompted whether you also want to delete all the data, or you can specify this with the --delete-data or --keep-data flags.

Changes in process launch functions#

The aiida.engine.submit method now accepts the argument wait. When set to True, instead of returning the process node straight away, the function will wait for the process to terminate before returning. By default it is set to False so the current behavior remains unchanged.

from aiida.engine import submit
node = submit(Process, wait=True)  # This call will block until process is terminated
assert node.is_terminated

This new feature is mostly useful for interactive demos and tutorials in notebooks. In these situations, it might be beneficial to use aiida.engine.run because the cell will be blocking until it is finished, indicating to the user that something is processing. When using submit, the cell returns immediately, but the results are not ready yet and typically the next cell cannot yet be executed. Instead, the demo should redirect the user to using something like verdi process list to query the status of the process.

However, using run has downsides as well, most notably that the process will be lost if the notebook gets disconnected. For processes that are expected to run longer, this can be really problematic, and so submit will have to be used regardless. With the new wait argument, submit provides the best of both worlds.

Although very useful, the introduction of this feature does break any processes that define wait or wait_interval as an input. Since the inputs to a process are defined as keyword arguments, these inputs would overlap with the arguments to the submit method. To solve this problem, inputs can now also be passed as a dictionary, e.g., where one would do before:

submit(SomeProcess, x=Int(1), y=Int(2), code=load_code('some-code'))
# or alternatively
inputs = {
    'x': Int(1),
    'y': Int(2),
    'code': load_code('some-code'),
}
submit(SomeProcess, **inputs)

The new syntax allows the following:

inputs = {
    'x': Int(1),
    'y': Int(2),
    'code': load_code('some-code'),
}
submit(SomeProcess, inputs)

Passing inputs as keyword arguments is still supported because sometimes that notation is still more legible than defining an intermediate dictionary. However, if both an input dictionary and keyword arguments are define, an exception is raised.

Improvements for built-in data types#

The XyData and ArrayData data plugins now allow to directly pass the content in the constructor. This allows defining the complete node in a single line

import numpy as np
from aiida.orm import ArrayData, XyData

xy = XyData(np.array([1, 2]), np.array([3, 4]), x_name='x', x_units='E', y_names='y', y_units='F')
assert all(xy.get_x()[1] == np.array([1, 2]))

array = ArrayData({'a': np.array([1, 2]), 'b': np.array([3, 4])})
assert all(array.get_array('a') == np.array([1, 2]))

It is now also no longer required to specify the name in ArrayData.get_array as long as the node contains just a single array:

import numpy as np
from aiida.orm import ArrayData

array = ArrayData(np.array([1, 2]))
assert all(array.get_array() == np.array([1, 2]))

Repository interface improvements#

As of v2.0.0, the repository interface of the Node class was moved to the Node.base.repository namespace. This was done to clean up the top-level namespace of the Node class which was getting very crowded, and in most use-cases, a user never needs to directly access these methods. It is up to the data plugin to provide specific methods to retrieve data that might be stored in the repository. For example, with the ArrayData, a user should now have to go to ArrayData.base.repository.get_object_content to retrieve an array from the repository, but the class provides ArrayData.get_array as a shortcut.

A few data plugins that ship with aiida-core didn’t respect this guideline, most notably the FolderData and SinglefileData plugins. This has been corrected in this release: for FolderData, all the repository methods are now once again directly available on the top-level namespace. The SinglefileData now makes it easier to get the content as bytes. Before, one had to do:

from aiida.orm import SinglefileData
node = SinglefileData.from_string('some content')
with node.open(mode='rb') as handle:
    byte_content = handle.read()

this can now be achieved with:

from aiida.orm import SinglefileData
node = SinglefileData.from_string('some content')
byte_content = node.get_content(mode='rb')

As of v2.0, due to the repository redesign, it was no longer possible to access a file directly by a filepath on disk. The repository interface only interacts with file-like objects to stream the content. However, a lot of Python libraries expect filepaths on disk and do not support file-like objects. This would force an AiiDA user to write the file from the repository to a temporary file on disk, and pass that temporary filepath. For example, consider the numpy.loadtxt function which requires a filepath, the code would look something like:

import pathlib
import shutil
import tempfile

with tempfile.TemporaryDirectory() as tmp_path:

    # Copy the entire content to the temporary folder
    dirpath = pathlib.Path(tmp_path)
    node.base.repository.copy_tree(dirpath)

    # Or copy the content of a file. Should use streaming
    # to avoid reading everything into memory
    filepath = (dirpath / 'some_file.txt')
    with filepath.open('rb') as target:
        with node.base.repository.open('rb') as source:
            shutil.copyfileobj(source, target)

    # Now use `filepath` to library call, e.g.
    numpy.loadtxt(filepath)

This burdensome boilerplate has now been made obsolete by the as_path method:

with node.base.repository.as_path() as filepath:
    numpy.loadtxt(filepath)

For the FolderData and SinglefileData plugins, the method can be accessed on the top-level namespace of course.

Full list of changes#

Features#

  • Add the SqliteDosStorage storage backend [702f88788]

  • XyData: Allow defining array(s) on construction [f11598dc6]

  • ArrayData: Make name optional in get_array [7fbe67cb6]

  • ArrayData: Allow defining array(s) on construction [35e669fe8]

  • FolderData: Expose repository API on top-level namespace [3e1f87373]

  • Repository: Add the as_path context manager [b0546e8ed]

  • Caching: Add the strict argument configuration validation [f272e197e]

  • Caching: Try to import an identifier if it is a class path [2c56fc234]

  • CLI: Add the command verdi profile setup [351021164]

  • CLI: Add cached and cached_from projections to verdi process list [3b445c4f1]

  • CLI: Add --all flag to verdi process kill [db1375949]

  • CLI: Lazily validate entry points in parameter types [d3807d422]

  • CLI: Add repair hint to verdi process play/pause/kill [8bc31bfd1]

  • CLI: Add the verdi process repair command [3e3d9b9f7]

  • CLI: Validate strict in verdi config set caching.disabled_for [9cff59232]

  • DynamicEntryPointCommandGroup: Allow entry points to be excluded [9e30ec8ba]

  • Add the aiida.common.log.capture_logging utility [9006eef3a]

  • Config: Add the create_profile method [ae7abe8a6]

  • Engine: Add the await_processes utility function [45767f050]

  • Engine: Add the wait argument to submit [8f5e929d1]

  • ORM: Add the User.is_default property [a43c4cd0f]

  • ORM: Add NodeCaching.CACHED_FROM_KEY for _aiida_cached_from constant [35fc3ae57]

  • ORM: Add the Entity.get_collection classmethod [305f1dbf4]

  • ORM: Add the Dict.get method [184fcd16e]

  • ORM: Register numpy.ndarray with the to_aiida_type to ArrayData [d8dd776a6]

  • Manager: Add the set_default_user_email [8f8f55807]

  • CalcJob: Add support for nested targets in remote_symlink_list [0ec650c1a]

  • RemoteData: Add the is_cleaned property [2a2353d3d]

  • SqliteTempBackend: Add support for reading from and writing to archives [83fc5cf69]

  • StorageBackend: Add the read_only class attribute [8a4303ff5]

  • SinglefileData: Add mode keyword to get_content [d082df7f1]

  • BaseRestartWorkChain: Factor out attachment of outputs [d6093d101]

  • Add support for NodeLinksManager to YAML serializer [6905c134e]

Performance#

  • CLI: Make loading of config lazy for improved responsiveness [d533b7a54]

  • Cache the lookup of entry points [12cc930db]

  • Refactor: Delay import of heavy packages to speed up import time [5dda6fd97]

  • Refactor: Delay import of heavy packages to speed up import time [8e6e08dc7]

  • Do not import aiida.cmdline in aiida.orm [0879a4e27]

  • Lazily define __type_string in orm.Group [ebf3101d9]

  • Lazily define _plugin_type_string and _query_type_string of Node` [3a61a7003]

Changes#

  • CLI: verdi profile delete is now storage plugin agnostic [5015f5fe1]

  • CLI: Usability improvements for interactive verdi setup [c53ea20a4]

  • CLI: Do not load config in defaults and callbacks during tab-completion [062058862]

  • Engine: Make process inputs in launchers positional [6d18ccb86]

  • Remove aiida.manage.configuration.load_documentation_profile [9941266ce]

  • ORM: Sealable.seal() return self instead of None [16e3bd3b5]

  • ORM: Move deprecation warnings from module level [c4afdb9be]

  • Config: Switch from jsonschema to pydantic [4203f162d]

  • DynamicEntryPointCommandGroup: Use pydantic to define config model [1d8ea2a27]

  • Config: Remove use of NO_DEFAULT for Option.default [275718cc8]

Fixes#

  • Add the report method to logging.LoggerAdapter [7d6684ce1]

  • CalcJob: Fix MPI behavior if withmpi option default is True [84737506e]

  • CalcJobNode: Fix validation for depth=None in retrieve_list [03c86d5c9]

  • CLI: Fix bug in verdi data core.trajectory show for various formats [fd4c1269b]

  • CLI: Add missing entry point groups for verdi plugin list [ae637d8c4]

  • CLI: Remove loading backend for verdi plugin list [34e564ad0]

  • CLI: Fix repository being required for verdi quicksetup [d4666009e]

  • CLI: Fix verdi config set when setting list option [314917801]

  • CLI: Keep list unique in verdi config set --append [3844f86c6]

  • CLI: Improve the formatting of verdi user list [806d7e236]

  • CLI: Set defaults for user details in profile setup [8b8887e55]

  • CLI: Reuse options in verdi user configure from setup [1c0b702ba]

  • InteractiveOption: Fix validation being skipped if ! provided [c4b183bc6]

  • ORM: Fix problem with detached DbAuthInfo instances [ec2c6a8fe]

  • ORM: Check nodes are from same backend in validate_link [7bd546ebe]

  • ORM: ProcessNode.is_valid_cache is False for unsealed nodes [a1f456d43]

  • ORM: Explicitly pass backend when constructing new entity [96667c8c6]

  • ORM: Replace .collection(backend) with .get_collection(backend) [bac2152c4]

  • Make warn_deprecation respect the warnings.showdeprecations option [6c28c63e9]

  • PsqlDosBackend: Fix changes not persisted after iterall and iterdict [2ea5087c0]

  • PsqlDosBackend: Fix Node.store excepting when inside a transaction [624dcd9fc]

  • Parser.parse_from_node: Validate outputs against process spec [d16792f3d]

  • Fix QueryBuilder.count for storage backends using sqlite [5dc1555bc]

  • Process functions: Fix bug with variable arguments [ca8bbc67f]

  • SqliteZipBackend: Return self in store [6a43b3f15]

  • SqliteZipBackend: Ensure the filepath is absolute and exists [5eac8b49d]

  • Remove with_dbenv use in aiida.orm [35c57b9eb]

Deprecations#

  • Deprecated aiida.orm.nodes.data.upf and verdi data core.upf [6625fd245]

Documentation#

  • Add topic section on storage [83dbe1ad9]

  • Add important note on using iterall and iterdict [0aea7e41b]

  • Add links about “entry point” and “plugin” to tutorial [517ffcb1c]

  • Disable the warnings.showdeprecations option [4adb06c0c]

  • Fix instructions for inspecting archive files [0a9c2788e]

  • Changes are reverted if exception during iterall [17c5d8724]

  • Various minor fixes to run_docker.rst [d3788adea]

  • Update pydata-sphinx-theme and add Discourse links [13df42c14]

  • Correct example of verdi config unset in troubleshooting [d6143dbc8]

  • Improvements to sections containing recently added functionality [836419f66]

  • Fix typo in run_codes.rst [9bde86ec7]

  • Fixtures: Fix suppress_warnings of run_cli_command [9807cede4]

  • Update citation suggestions [1dafdf2dd]

Dependencies#

Devops#

  • CI: Use Python 3.10 for pre-commit in CI and CD workflows [f41c8ac90]

  • CI: Using concurrency for CI actions [4db54b7f8]

  • CI: Update tox to use Python 3.9 [227390a52]

  • Docker: Bump upload-artifact action to v4 for Docker workflow [bfdb2828a]

  • Refactor: Replace all with iterall where beneficial [8a2fece02]

  • Pre-commit: Disable no-member and no-name-in-module for aiida.orm [15379bbee]

  • Tests: Move memory leak tests to main unit test suite [561f93cef]

  • Tests: Move ipython magic tests to main unit test suite [ce9acc312]

  • Tests: Remove deprecated aiida/manage/tests/main module [5b9da7d1e]

  • Tests: Refactor transport tests from unittest to pytest [ec64780c2]

  • Tests: Fix failing tests/cmdline/commands/test_setup.py [b6f7ec188]

  • Tests: Print stack trace if CLI command excepts with run_cli_command [08cba0f78]

  • Tests: Make PsqlDosStorage profile unload test more robust [1c72eac1f]

  • Tests: Fix flaky work chain tests using recwarn fixture [207151784]

  • Tests: Fix StructureData test breaking for recent pymatgen versions [d1d64e800]

  • Typing: Improve annotations of process functions [a85af4f0c]

  • Typing: Add type hinting for aiida.orm.nodes.data.array.xy [2eaa5449b]

  • Typing: Add type hinting for aiida.orm.nodes.data.array.array [c19b1423a]

  • Typing: Add overload signatures for open [0986f6b59]

  • Typing: Add overload signatures for get_object_content [d18eedc8b]

  • Typing: Correct type annotation of WorkChain.on_wait [923cc314c]

  • Typing: Improve type hinting for aiida.orm.nodes.data.singlefile [b9d087dd4]

v2.4.2 - 2023-11-30#

Docker#

  • Disable the consumer timeout for RabbitMQ [5ce1e7ec3]

  • Add rsync and graphviz to system requirements [c4799add4]

Dependencies#

v2.4.1 - 2023-11-15#

This patch release comes with an improved set of Docker images and a few fixes to provide compatibility with recent versions of pymatgen.

Docker#

  • Improved Docker images [fec4e3bc4]

  • Add folders that automatically run scripts before/after daemon start in Docker image [fe4bc1d3d]

  • Pass environment variable to aiida-prepare script in Docker image [ea47668ea]

  • Update the .devcontainer to use the new docker stack [413a0db65]

Dependencies#

  • Add compatibility for pymatgen>=v2023.9.2 [1f6027f06]

Devops#

  • Tests: Make PsqlDosStorage profile unload test more robust [f392459bd]

  • Tests: Fix StructureData test breaking for recent pymatgen versions [093037d48]

  • Trigger Docker image build when pushing to support/* branch [5cf3d1d75]

  • Use aiida-core-base image from ghcr.io [0e5b1c747]

  • Loosen trigger conditions for Docker build CI workflow [22e8a8069]

  • Follow-up docker build runner macOS-ARM64 [1bd9bf03d]

  • Upload artifact by PR from forks for docker workflow [afc2dad8a]

  • Update the image name for docker image [17507b410]

v2.4.0 - 2023-06-22#

This minor release comes with a number of new features and improvements as well as a significant amount of bug fixes. Support for Python 3.8 has been officially dropped in accordance with AEP 003.

As a result of one of the bug fixes, related to the caching of CalcJob nodes, a database migration had to be added, the first since the release of v2.0. After ugrading to v2.4.0, you will be prompted to migrate your database. The automated migration drops the hashes of existing CalcJobNodes and provides you with the optional command to recompute them. Execute the command if existing CalcJobNodes need to be usable as valid cache sources.

Features#

  • Config: Add option to change recursion limit in daemon workers [226159fd9]

  • CLI: Added compress option to verdi storage maintain [add474cbb]

  • Expose get_daemon_client so it can be imported from aiida.engine [1a0c1ee93]

  • verdi computer test: Improve messaging of login shell check [062a58260]

  • verdi node rehash: Add aiida.node as group for --entry-point [2fd07514d]

  • verdi process status: Add call_link_label to stack entries [bd9372a5f]

  • SinglefileData: Add the from_string classmethod [c25de615e]

  • DynamicEntryPointCommandGroup: Add support for shared options [220a65c76]

  • DynamicEntryPointCommandGroup: Pass ctx to command callable [7de711be4]

  • ProcessNode: Add the exit_code property [ad8a539ee]

Fixes#

  • Engine: Dynamically update maximum stack size close to overflow to address RecursionError under heavy load [f797b4766]

  • CalcJobNode: Fix the computation of the hash [685e0f87d]

  • CalcJob: Ignore file in remote_copy_list not existing [101a8d61b]

  • CalcJob: Assign outputs from node in case of cache hit [777b97601]

  • Fix log messages being logged twice to the daemon log file [bfd63c790]

  • Process control: Change language when not waiting for response [68cb4579d]

  • Do not assume pgtest cluster started in postgres_cluster fixture [1de2ca576]

  • Process control: Warn instead of except when daemon is not running [ad4fbcccb]

  • DirectScheduler: Add ? as JobState.UNDETERMINED [ffc869d8f]

  • CLI: Correct verdi devel rabbitmq tasks revive docstring [13cadd05f]

  • SinglefileData: Fix bug when filename is pathlib.Path [f36bf583c]

  • Improve clarity of various deprecation warnings [c72a252ed]

  • CalcJob: Remove default of withmpi input and make it optional [6a88cb315]

  • Process: Have inputs property always return AttributesFrozenDict [60756fe30]

  • PsqlDos: Add migration to remove hashes for all CalcJobNodes [7ad916836]

  • PsqlDosMigrator: Commit changes when migrating existing schema [f84fe5b60]

  • PsqlDos: Add entry_point_string argument to drop_hashes [c7a36fa3d]

  • PsqlDos: Make hash reset migrations more explicit [c447a1af3]

  • verdi process list: Fix double percent sign in daemon usage [68be866e6]

  • Fix the daemon_client fixture [9e5f5eefd]

  • Transports: Raise FileNotFoundError in copy if source doesn’t exist [d82069441]

Devops#

  • Add graphviz to system requirements of RTD build runner [3df02550e]

  • Add types for DefaultFieldsAttributeDict subclasses [afed5dc46]

  • Bump Python version for RTD build [5df446cd3]

  • Pre-commit: Fix mypy warning in aiida.orm.utils.serialize [c25922484]

  • Update Docker base image aiida-prerequisites==0.7.0 [ac755afae]

  • Use f-strings in aiida/engine/daemon/execmanager.py [49cffff21]

Dependencies#

Deprecations#

  • QueryBuilder: Deprecate debug argument and use logger [603ff37a0]

Documentation#

  • Add missing core. prefix to all verdi data subcommands [99319b3c1]

  • Clarify negation operator in QueryBuilder filters [2c828811f]

  • Correct “variable” to “variadic” arguments [978217693]

  • Fix reference target warnings related to flask_restful [4f76e0bd7]

v2.3.1 - 2023-05-22#

Fixes#

  • DaemonClient: Clean stale PID file in stop_daemon [#6007]

v2.3.0 - 2023-04-17#

This release comes with a number of improvements, some of the more useful and important of which are quickly highlighted. A full list of changes can be found below.

Process function improvements#

A number of improvements in the usage of process functions, i.e., calcfunction and workfunction, have been added. Each subsection title is a link to the documentation for more details.

Variadic arguments#

Variadic arguments can be used in case the function should accept a list of inputs of unknown length. Consider the example of a calculation function that computes the average of a number of Int nodes:

@calcfunction
def average(*args):
    return sum(args) / len(args)

result = average(*(1, 2, 3))

Automatic type validation#

Type hint annotations can now be used to add automatic type validation to process functions.

@calcfunction
def add(x: Int, y: Int):
    return x + y

add(1, 1.0)  # Passes
add(1, '1.0')  # Raises an exception

Since the Python base types (int, str, bool, etc.) are automatically serialized, these can also be used in type hints. The following example is therefore identical to the previous:

@calcfunction
def add(x: int, y: int):
    return x + y

Docstring parsing#

The calcfunction and workfunction generate a Process of the decorated function on-the-fly. In doing so, it automatically defines the ProcessSpec that is normally done manually, such as for a CalcJob or a WorkChain. Before, this would just define the ports that the function process accepts, but the help attribute of the port would be left empty. This is now parsed from the docstring, if it can be correctly parsed:

@calcfunction
def add(x: int, y: int):
    """Add two integers.

    :param x: Left hand operand.
    :param y: Right hand operand.
    """
    return x + y

assert add.spec().inputs['a'].help == 'Left hand operand.'
assert add.spec().inputs['b'].help == 'Right hand operand.'

This functionality is particularly useful when exposing process functions in work chains. Since the process specification of the exposed function will be automatically inherited, the user can inspect the help string through the builder. The automatic documentation produced by the Sphinx plugin will now also display the help string parsed from the docstring.

Nested labels for output nodes#

The keys in the output dictionary can now contain nested namespaces:

@calcfunction
def add(alpha, beta):
    return {'nested.sum': alpha + beta}

result = add(Int(1), Int(2))
assert result['nested']['sum'] == 3

As class member methods#

Process functions can now be defined as class member methods of work chains:

class CalcFunctionWorkChain(WorkChain):

    @classmethod
    def define(cls, spec):
        super().define(spec)
        spec.input('x')
        spec.input('y')
        spec.output('sum')
        spec.outline(
            cls.run_compute_sum,
        )

    @staticmethod
    @calcfunction
    def compute_sum(x, y):
        return x + y

    def run_compute_sum(self):
        self.out('sum', self.compute_sum(self.inputs.x, self.inputs.y))

The function should be declared as a staticmethod and it should not include the self argument in its function signature. It can then be called from within the work chain as self.function_name(*args, **kwargs).

Scheduler plugins: including environment_variables#

The Scheduler base class implements the concrete method _get_submit_script_environment_variables which formats the lines for the submission script that set the environment variables that were defined in the metadata.options.environment_variables input. Before it was left up to the plugins to actually call this method in the _get_submit_script_header, but this is now done by the base class in the get_submit_script. You can now remove the call to _get_submit_script_environment_variables from your scheduler plugins, as the base class will take care of it. A deprecation warning is emitted if the base class detects that the plugin is still calling it manually. See the pull request for more details.

WorkChain: conditional predicates should return boolean-like#

Up till now, work chain methods that are used as the predicate in a conditional, e.g., if_ or while_ could return any type. For example:

class SomeWorkChain(WorkChain):

    @classmethod
    def define(cls, spec):
        super().define(spec)
        spec.outline(if_(cls.some_conditional)())

    def some_conditional(self):
        if self.ctx.something == 'something':
            return True

The some_conditional method is used as the “predicate” of the if_ conditional. It returns True or None. Since the None value in Python is “falsey”, it would be considered as returning False. However, this duck-typing could accidentally lead to unexpected situations, so we decided to be more strict on the return type. As of now, a deprecation warning is emitted if the method returns anything that is not “boolean-like”, i.e., does not implement the __bool__ method. If you see this warning, please make sure to return a boolean, like the built-ins True or False, or a numpy.bool or aiida.orm.Bool. See the pull request for more details.

Controlling usage of MPI#

It is now possible to define on a code object whether it should be run with or without MPI through the with_mpi attribute. It can be set from the Python API as AbstractCode(with_mpi=with_mpi) or through the --with-mpi / --no-with-mpi option of the verdi code create CLI command. This option adds a manner to control the use of MPI in calculation jobs, in addition to the existing ones defined by the CalcJob plugin and the metadata.options.withmpi input. For more details on how these are controlled and how conflicts are handled, please refer to the documentation.

Add support for Docker containers#

Support is added for running calculation within Docker containers. For example, to run Quantum ESPRESSO pw.x in a Docker container, write the following file to config.yml:

label: qe-pw-on-docker
computer: localhost
engine_command: docker run -i -v $PWD:/workdir:rw -w /workdir {image_name} sh -c
image_name: haya4kun/quantum_espresso
filepath_executable: pw.x
default_calc_job_plugin: quantumespresso.pw
use_double_quotes: false
wrap_cmdline_params: true

and run the CLI command:

verdi code create core.code.containerized --config config.yml --non-interactive

This should create a ContainerizedCode that you can now use to launch a PwCalculation. For more details, please refer to the documentation.

Exporting code configurations#

It is now possible to export the configuration of an existing code through the verdi code export command. The produced YAML file can be used to recreate the code through the verdi code create command. Note that you should use the correct subcommand based on the type of the original code. For example, if it was an InstalledCode you should use verdi code create core.code.installed. For the legacy Code instances, you should use verdi code setup. See the pull request for more details.

Full list of changes#

Features#

  • AbstractCode: Add the with_mpi attribute [#5922]

  • ContainerizedCode: Add support for Docker images to use as Code for CalcJobs [#5841]

  • InstalledCode: Allow relative path for filepath_executable [#5879]

  • CLI: Allow specifying output filename in verdi node graph generate [#5897]

  • CLI: Add --timeout option to all verdi daemon commands [#5966]

  • CLI: Add the verdi calcjob remotecat command [#4861]

  • CLI: Add the verdi code export command [#5860]

  • CLI: Improved customizability and scriptability of verdi storage maintain [#5936]

  • CLI: verdi quicksetup: Further reduce required user interaction [#5768]

  • CLI: verdi computer test: Add test for login shell being slow [#5845]

  • CLI: verdi process list: Add exit_message as projectable attribute [#5853]

  • CLI: verdi node delete: Add verbose list of pks to be deleted [#5878]

  • CLI: Fail command if --config file contains unknown key [#5939]

  • CLI: verdi daemon status: Do not except when no profiles are defined [#5874]

  • ORM: Add unary operations +, - and abs to NumericType [#5946]

  • Process functions: Support class member functions as process functions [#4963]

  • Process functions: Infer argument valid_type from type hints [#5900]

  • Process functions: Parse docstring to set input port help attribute [#5919]

  • Process functions: Add support for variadic arguments [#5691]

  • Process functions: Allow nested output namespaces [#5954]

  • Process: Store JSON-serializable metadata inputs on the node [#5801]

  • Port: Add the is_metadata keyword [#5801]

  • ProcessBuilder: Include metadata inputs in get_builder_restart [#5801]

  • StructureData: Add mode argument to get_composition [#5926]

  • Scheduler: Allow terminating job if submission script is invalid [#5849]

  • SlurmScheduler: Detect broken submission scripts for invalid account [#5850]

  • SlurmScheduler: Parse the NODE_FAIL state [#5866]

  • WorkChain: Add dataclass serialisation to context [#5833]

  • IcsdDbImporter: Add is_theoretical tag to queried entries [#5868]

Fixes#

  • CLI: Prefix the verdi data subcommands with core. [#5846]

  • CLI: Respect config log levels if --verbosity not explicitly passed [#5925]

  • CLI: verdi config list: Do not except if no profiles are defined [#5921]

  • CLI: verdi code show: Add missing code attributes [#5916]

  • CLI: verdi quicksetup: Fix error incorrect role when creating database [#5828]

  • CLI: Fix error in aiida.cmdline.utils.log.CliFormatter [#5957]

  • Daemon: Fix false-positive of stopped daemon in verdi daemon status [#5862]

  • DaemonClient: Fix and homogenize use of timeout in client calls [#5960]

  • ProcessBuilder: Fix bug in _recursive_merge [#5801]

  • QueryBuilder: Catch new exception raised by sqlalchemy>=1.4.45 [#5875]

  • Fix the %verdi IPython magics utility [#5961]

  • Fix bug in aiida.engine.utils.instantiate_process [#5952]

  • Fix incorrect import of exception from kiwipy.communications [#5947]

Deprecations#

  • Scheduler: Move setting of environment variables into base class [#5948]

  • WorkChains: Emit deprecation warning if predicate if_/while_ does not return boolean-like [#5924]

Changes#

  • DaemonClient: Refactor to include parsing of client response [#5850]

  • ORM: Remove Entity.from_backend_entity from the public API [#5447]

  • PbsproScheduler: Replace deprecated ppn tag with ncpus [#5910]

  • ProcessBuilder: Move _prune method to standalone utility [#5801]

  • verdi process list: Simplify the daemon load implementation [#5850]

Documentation#

  • Add FAQ on MFA-enabled computers [#5887]

  • Add link to all metadata.options inputs in CalcJob submission example [#5912]

  • Add warning that Data constructor is not called on loading [#5898]

  • Add note on how to create a code that uses Conda environment [#5905]

  • Add --without-daemon flag to benchmark script [#5839]

  • Add alternative for conda env activation in submission script [#5950]

  • Clarify that process functions can be exposed in work chains [#5919]

  • Fix the intro/tutorial.md notebook [#5961]

  • Fix the overindentation of lists [#5915]

  • Hide the “Edit this page” button on the API reference pages [#5956]

  • Note that an entry point is required for using a data plugin [#5907]

  • Set use_login_shell=False for localhost in performance benchmark [#5847]

  • Small improvements to the benchmark script [#5854]

  • Use mamba instead of conda [#5891]

DevOps#

  • Add devcontainer for easy integration with VSCode [#5913]

  • CI: Update sphinx-intl and install transifex CLI [#5908]

  • Fix the test-install workflow [#5873]

  • Pre-commit: Improve typing of aiida.schedulers.scheduler [#5849]

  • Pre-commit: Set yapf option allow_split_before_dict_value = false[#5931]

  • Process functions: Replace getfullargspec with signature [#5900]

  • Fixtures: Add argument use_subprocess to run_cli_command [#5846]

  • Fixtures: Change default use_subprocess=False for run_cli_command [#5846]

  • Tests: Use use_subprocess=False and suppress_warnings=True [#5846]

  • Tests: Fix bugs revealed by running with use_subprocess=True [#5846]

  • Typing: Annotate aiida/orm/utils/serialize.py [#5832]

  • Typing: Annotate aiida/tools/visualization/graph.py [#5821]

  • Typing: Use modern syntax for aiida.engine.processes.functions [#5900]

Dependencies#

  • Add compatibility for ipython~=8.0 [#5888]

  • Bump cryptography from 36.0.0 to 39.0.1 [#5885]

  • Remove upper limit on werkzeug [#5904]

  • Update pre-commit requirement isort==5.12.0 [#5877]

  • Update requirement importlib-metadata~=4.13 [#5963]

  • Bump graphviz version to 0.19 [#5965]

New contributors#

Thanks a lot to the following new contributors:

v2.2.2 - 2023-02-10#

Fixes#

  • Critical bug fix: Fix bug causing CalcJobs to except after restarting daemon [#5886]

v2.2.1 - 2022-12-22#

Fixes#

  • Critical bug fix: Revert the changes of PR [#5804] released with v2.2.0, which addressed a bug when mutating nodes during QueryBuilder.iterall. Unfortunately, the change caused changes performed by verdi commands (as well as changes made in verdi shell) to not be persisted to the database. [#5851]

v2.2.0 - 2022-12-13#

This feature release comes with a significant feature and a number of improvements and fixes.

Live calculation job monitoring#

In certain use cases, it is useful to have a calculation job stopped prematurely, before it finished or the requested wallclock time runs out. Examples are calculations that seem to be going nowhere and so continuing would only waste computational resources. Up till now, a calculation job could only be “manually” stopped, through verdi process kill. In this release, functionality is added that allows calculation jobs to be monitored automatically by the daemon and have them stopped when certain conditions are met.

Monitors can be attached to a calculation job through the monitors input namespace:

builder = load_code().get_builder()
builder.monitors = {
    'monitor_a': Dict({'entry_point': 'some.monitor'}),
    'monitor_b': Dict({'entry_point': 'some.other.monitor'}),
}

Monitors are referenced by their entry points with which they are registered in the aiida.calculations.monitors entry point group. A monitor is essentially a function that implements the following interface:

from aiida.orm import CalcJobNode
from aiida.transports import Transport

def monitor(node: CalcJobNode, transport: Transport) -> str | CalcJobMonitorResult | None:
    """Retrieve and inspect files in working directory of job to determine whether the job should be killed.

    :param node: The node representing the calculation job.
    :param transport: The transport that can be used to retrieve files from remote working directory.
    :returns: A string if the job should be killed, `None` otherwise.
    """

The transport allows to fetch files from the working directory of the calculation. If the job should be killed, the monitor simply returns a string with the message why and the daemon will send the message to kill the job.

For more information and a complete description of the interface, please refer to the documentation. This functionality was accepted based on AEP 008 which provides more detail on the design choices behind this implementation.

Full list of changes#

Features#

  • CalcJob: Add functionality that allows live monitoring [#5659]

  • CLI: Add --raw option to verdi code list [#5763]

  • CLI: Add the -h short-hand flag for --help to verdi [#5792]

  • CLI: Add short option names for verdi code create [#5799]

  • StorageBackend: Add the initialise method [#5760]

  • Fixtures: Add support for Process inputs to submit_and_await [#5780]

  • Fixtures: Add aiida_computer_local and aiida_computer_ssh [#5786]

  • Fixtures: Modularize fixtures creating AiiDA test instance and profile [#5758]

  • Computer: Add the is_configured property [#5786]

  • Plugins: Add aiida.storage to ENTRY_POINT_GROUP_FACTORY_MAPPING [#5798]

Fixes#

  • verdi run: Do not add pathlib.Path instance to sys.path [#5810]

  • Process functions: Restore support for dynamic nested input namespaces [#5808]

  • Process: properly cleanup when exception in state transition [#5697]

  • Process: Update outputs before updating node process state [#5813]

  • PsqlDosMigrator: refactor the connection handling [#5783]

  • PsqlDosBackend: Use transaction whenever mutating session state, fixing exception when storing a node or group during QueryBuilder.iterall [#5804]

  • InstalledCode: Fix bug in validate_filepath_executable for SSH [#5787]

  • WorkChain: Protect public methods from being subclassed. Now if you accidentally override, for example, the run method of the WorkChain, an exception is raised instead of silently breaking the work chain [#5779]

Changes#

  • Rename PsqlDostoreMigrator to PsqlDosMigrator [#5761]

  • ORM: Remove pymatgen version check in StructureData.set_pymatgen_structure [#5777]

  • StorageBackend: Remove recreate_user from _clear [#5772]

  • PsqlDosMigrator: Remove hardcoding of table name in database reset [#5781]

Dependencies#

  • Dependencies: Add support for Python 3.11 [#5778]

Documentation#

  • Docs: Correct command to enable verdi tab-completion for fish shell [#5784]

  • Docs: Fix transport & scheduler type in localhost setup [#5785]

  • Docs: Fix minor formatting issues in “How to run a code” [#5794]

DevOps#

  • CI: Increase load limit for verdi to 0.5 seconds [#5773]

  • CI: Add workflow_dispatch trigger to nightly.yml [#5760]

  • ORM: Fix typing of aiida.orm.nodes.data.code module [#5830]

  • Pin version of setuptools as it breaks dependencies [#5782]

  • Tests: Use explicit aiida_profile_clean in process control tests [#5778]

  • Tests: Replace all use of aiida_profile_clean with aiida_profile where a clean profile is not necessary [#5814]

  • Tests: Deal with run_via_daemon returning None in RPN tests [#5813]

  • Make type-checking opt-out [#5811]

v2.1.2 - 2022-11-14#

Fixes#

  • BaseRestartWorkChain: Fix bug in _wrap_bare_dict_inputs introduced in v2.1.0 [#5757]

v2.1.1 - 2022-11-10#

Fixes#

  • Engine: Remove *args from the Process.submit method. [#5753] Positional arguments were silently ignored leading to a misleading error message. For example, if a user called

    inputs = {}
self.submit(cls, inputs)

    instead of the intended

    inputs = {}
self.submit(cls, **inputs)

    The returned error message was that one of the required inputs was not defined. Now it will correctly raise a TypeError saying that positional arguments are not supported.

  • Process functions: Add serialization for Python base type defaults [#5744] Defining Python base types as defaults, such as:

    @calcfunction
def function(a, b = 5):
    return a + b

    would raise an exception. The default is now automatically serialized, just as an input argument would be upon function call.

  • Process control: Reinstate process status for paused/killed processes [#5754] Regression introduced in aiida-core==2.1.0 caused the message Killed through 'verdi process list' to no longer be set on the process_status of the node.

  • QueryBuilder: use a nested session in iterall and iterdict [#5736] Modifying entities yielded by QueryBuilder.iterall and QueryBuilder.iterdict would raise an exception, for example:

    for [node] in QueryBuilder().append(Node).iterall():
    node.base.extras.set('some', 'extra')

v2.1.0 - 2022-11-07#

This feature release comes with a number of new features as well as quite a few fixes of bugs and stability issues. Further down you will find a complete list of changes, after a short description of some of the most important changes:

Automatic input serialization in calculation and work functions#

The inputs to calcfunctions and workfunctions are now automatically converted to AiiDA data types if they are one of the basic Python types (bool, dict, Enum, float, int, list or str). This means that code that looked like:

from aiida.engine import calcfunction
from aiida.orm import Bool, Float, Int, Str

@calcfunction
def function(switch, threshold, count, label):
    ...

function(Bool(True), Float(0.25), Int(10), Str('some-label'))

can now be simplified to:

from aiida.engine import calcfunction
from aiida.orm import Bool, Float, Int, Str

@calcfunction
def function(switch, threshold, count, label):
    ...

function(True, 0.25, 10, 'some-label')

Improved interface for creating codes#

The Code data plugin was a single class that served two different types of codes: “remote” codes and “local” codes. These names “remote” and “local” have historically caused a lot of confusion. Likewise, using a single class Code for both implementations also has led to confusing interfaces.

To address this issue, the functionality has been split into two new classes InstalledCode and PortableCode, that replace the “remote” and “local” code, respectively. The installed code represents an executable binary that is already pre-installed on some compute resource. The portable code represents a code (executable plus any additional required files) that are stored in AiiDA’s storage and can be automatically transfered to any computer before being executed.

Creating a new instance of these new code types is easy:

from pathlib import Path
from aiida.orm import InstalledCode, PortableCode

installed_code = InstalledCode(
    label='installed-code',
    computer=load_computer('localhost'),
    filepath_executable='/usr/bin/bash'
)

portable_code = PortableCode(
    label='portable-code',
    filepath_files=Path('/some/path/code'),
    filepath_executable='executable.exe'
)

Codes can also be created through the new verdi command verdi code create. To specify the type of code to create, pass the corresponding entry point name as an argument. For example, to create a new installed code, invoke:

verdi code create core.code.installed

The options for each subcommand are automatically generated based on the code type, and so only options that are relevant to that code type will be prompted for.

The new code classes both subclass the aiida.orm.nodes.data.code.abstract.AbstractCode base class. This means that both InstalledCodes and PortableCodes can be used as the code input for CalcJobs without problems.

The old Code class remains supported for the time being as well, however, it is deprecated and will be remove at some point. The same goes for the verdi code setup command; please use verdi code create instead. Existing codes will be automatically migrated to either an InstalledCode or a PortableCode. It is strongly advised that you update any code that creates new codes to use these new plugin types.

Support for running code in containers#

Support is added to run calculation jobs inside a container. A containerized code can be setup through the CLI:

verdi code create core.code.containerized \
    --label containerized \
    --image-name docker://alpine:3 \
    --filepath-executable /bin/sh \
    --engine-command "singularity exec --bind $PWD:$PWD {image_name}"

as well as through the API:

from aiida.orm import ContainerizedCode, load_computer
code = ContainerizedCode(
    computer=load_computer('some-computer')
    filepath_executable='/bin/sh'
    image_name='docker://alpine:3',
    engine_command='singularity exec --bind $PWD:$PWD {image_name}'
).store()

In the example above we use the Singularity containerization technology. For more information on what containerization programs are supported and how to configure them, please refer to the documentation.

Control daemon and processes from the API#

Up till now, the daemon and live processes could only easily be controlled through verdi daemon and verdi process, respectively. In this release, modules are added to provide the same functionality through the Python API.

Daemon API#

The daemon can now be started and stopped through the DaemonClient which can be obtained through the get_daemon_client utility function:

from aiida.engine.daemon.client import get_daemon_client
client = get_daemon_client()

By default, this will give the daemon client for the current default profile. It is also possible to explicitly specify a profile:

client = get_daemon_client(profile='some-profile')

The daemon can be started and stopped through the client:

client.start_daemon()
assert client.is_daemon_running
client.stop_daemon(wait=True)

Process API#

The functionality of verdi process to play, pause and kill is now made available through the aiida.engine.process.control module. Processes can be played, paused or killed through the play_processes, pause_processes, and kill_processes, respectively. The processes to act upon are defined through their ProcessNode which can be loaded using load_node.

from aiida.engine.process import control

processes = [load_node(<PK1>), load_node(<PK2>)]

pause_processes(processes)  # Pause the processes
play_processes(processes)  # Play them again
kill_processes(processes)  # Kill the processes

Instead of specifying an explicit list of processes, the functions also take the all_entries keyword argument:

pause_processes(all_entries=True)  # Pause all running processes

REST API can serve multiple profiles#

Before, a single REST API could only serve data of a single profile at a time. This limitation has been removed and a single REST API instance can now serve data from all profiles of an AiiDA instance. To maintain backwards compatibility, the new functionality needs to be explicitly enabled through the configuration:

verdi config set rest_api.profile_switching True

After the REST API is restarted, it will now accept the profile query parameter, for example:

http://127.0.0.1:5000/api/v4/computers?profile=some-profile-name

If the specified is already loaded, the REST API functions exactly as without profile switching enabled. If another profile is specified, the REST API will first switch profiles before executing the request.

If the profile parameter is specified in a request and the REST API does not have profile switching enabled, a 400 response is returned.

Pluginable data storage backends#

Warning: this is beta functionality. It is now possible to implement custom storage backends to control where all data of an AiiDA profile is stored. To provide a data storage plugin, one should implement the aiida.orm.implementation.storage_backend.StorageBackend interface. The default implementation provided by aiida-core is the aiida.storage.psql_dos.backend.PsqlDosBackend which uses a PostgreSQL database for the provenance graph and a disk-objectstore container for repository files.

Storage backend plugins should be registered in the new entry point group aiida.storage. The default storage backend PsqlDosBackend has the core.psql_dos entry point name.

The storage backend to be used for a profile can be specified using the --db-backend option in verdi setup and verdi quicksetup. The entry point of the selected backend is stored in the storage.backend key of a profile configuration:

{
    "profiles": {
        "profile-name": {
            "PROFILE_UUID": "",
            "storage": {
                "backend": "core.psql_dos",
                "config": {}
            },
            "process_control": {},
            "default_user_email": "aiida@localhost",
            "test_profile": false
        },

}

At the moment, it is not quite clear if the abstract interface StorageBackend properly abstracts everything that is needed to implement any storage backend. For the time being then, it is advised to subclass the PsqlDosBackend and replace parts required for the use-case, such as just replacing the file repository implementation.

Full list of changes#

Features#

  • Process: Add hook to customize the process_label attribute [#5713]

  • Add the ContainerizedCode data plugin [#5667]

  • API: Add the aiida.engine.processes.control module [#5630]

  • PluginVersionProvider: Add support for entry point strings [#5662]

  • verdi setup: Add the --profile-uuid option [#5673]

  • Process control: Add the revive_processes method [#5677]

  • Process functions: Add the get_source_code_function method [#4554]

  • CLI: Improve the quality of verdi code list output [#5750]

  • CLI: Add the verdi devel revive command [#5677]

  • CLI: verdi process status --max-depth [#5727]

  • CLI: verdi setup/quicksetup store autofill user info early [#5729]

  • CLI: Add the devel launch-add command [#5733]

  • CLI: Make filename in verdi node repo cat optional for SinglefileData [#5747]

  • CLI: Add the verdi devel rabbitmq command group [#5718]

  • API: Add function to start the daemon [#5625]

  • BaseRestartWorkChain: add the get_outputs hook [#5618]

  • CalcJob: extend retrieve_list syntax with depth=None [#5651]

  • CalcJob: allow wildcards in stash.source_list paths [#5601]

  • Add global config option rest_api.profile_switching [#5054]

  • REST API: make the profile configurable as request parameter [#5054]

  • ProcessFunction: Automatically serialize Python base type inputs [#5688]

  • BaseRestartWorkChain: allow to override priority in handler_overrides [#5546]

  • ORM: add entry_point classproperty to Node and Group [#5437]

  • Add the aiida.storage entry point group [#5501]

  • Add the config option storage.sandbox [#5501]

  • Add the InstalledCode and PortableCode data plugins [#5510]

  • CLI: Add the verdi code create command group [#5510]

  • CLI: Add the DynamicEntryPointCommandGroup command group [#5510]

  • Add a client to connect to RabbitMQ Manamegement HTTP API [#5718]

  • LsfScheduler: add support for num_machines [#5153]

  • JobResource: add the accepts_default_memory_per_machine [#5642]

  • AbstractCode: add abstraction methods for command line parameters [#5664]

  • ArithmeticAddCalculation: Add the metadata.options.sleep input [#5663]

  • DaemonClient: add the get_env method [#5631]

  • Tests: Make daemon fixtures available to plugin packages [#5701]

  • verdi plugin list: Show which exit codes invalidate cache [#5710]

  • verdi plugin list: Show full help for input and output ports [#5711]

Fixes#

  • ArrayData: replace nan and inf with None when dumping to JSON [#5613]

  • Archive: add missing migration of transport entry points [#5604]

  • BaseRestartWorkChain: fix handler_overrides ignoring enabled=False [#5598]

  • CLI: allow setting options for config without profiles [#5544]

  • CLI: normalize use of colors [#5547]

  • Config: fix bug in downgrade past version 6 [#5528]

  • DaemonClient: close CircusClient after call [#5631]

  • Engine: Do not call serializer for None values [#5694]

  • Engine: Do not let DuplicateSubcriberError except a Process [#5715]

  • ORM: raise when trying to pickle instance of Entity [#5549]

  • ORM: Return None in get_function_source_code instead of excepting [#5730]

  • Fix get_entry_point not raising even for duplicate entry points [#5531]

  • Fix: reference to command in message for verdi storage maintain [#5558]

  • Fix: is_valid_cache setter for ProcessNodes [#5583]

  • Fix exception when importing an archive into a profile with many nodes [#5740]

  • Profile: make definition of daemon filepaths dynamic [#5631]

  • Fixtures: Fix bug in reset of empty_config fixture [#5717]

  • PsqlDosBackend: ensure sqla sessions are garbage-collected on close [#5728]

  • TrajectoryData: Fix bug in get_step_data [#5734]

  • ProfileManager: restart daemon in clear_profile [#5751]

Changes#

  • Mark relevant Process exit codes as invalidates_cache=True[#5709]

  • TemplatereplacerCalculation: Change exit codes to be in 300 range [#5709]

  • Add the prefix core. to all storage entry points [#5501]

  • CalcJob: Fully abstract interaction with AbstractCode in presubmit [#5666]

  • CLI: make label the default group list order in verdi group list [#5523]

  • Config: add migration to properly prefix storage backend [#5501]

  • Move query utils from aiida.cmdline to aiida.tools [#5630]

  • SandboxFolder: decouple the location from the profile [#5496]

  • TemplatereplacerDoublerParser: rename and generalize implementation [#5669]

  • Process: Allow None for input ports that are not required [#5722]

Dependencies#

  • RabbitMQ: Remove support for v3.5 and older [#5718]

  • Relax wrapt requirement [#5607]

  • Set upper limit werkzeug<2.2 [#5606]

  • Update requirement click~=8.1 [#5504]

Deprecations#

  • Deprecate Profile.repository_path [#5516]

  • Deprecate: verdi code setup and CodeBuilder [#5510]

  • Deprecate the method aiida.get_strict_version [#5512]

  • Remove use of legacy Code [#5510]

Documentation#

  • Add section on basic performance benchmark with automated benchmark script [#5724]

  • Add -U flag to PostgreSQL database backup command [#5550]

  • Clarify excepted and killed calculations are not cached [#5525]

  • Correct snippet for workchain context nested keys [#5551]

  • Plugin package setup add PEP 621 example [#5626]

  • Remove note on disk space for caching [#5534]

  • Remove explicit release tag in Docker image name [#5671]

  • Remove example REST API extension with POST requests [#5737]

  • Resubmit a Process from a ProcessNode [#5579]

Devops#

  • Add a notification for nightly workflow on fail [#5605]

  • CI: Remove --use-feature flag in pip install of CI [#5703]

  • Fixtures: Add started_daemon_client and stopped_daemon_client [#5631]

  • Fixtures: Add the entry_points fixture to dynamically add and remove entry points [#5745]

  • Refactor: Process extract CalcJob specific input handling from Process [#5539]

  • Refactor: remove unnecessary use of tempfile.mkdtemp [#5639]

  • Refactor: Remove internal use of various deprecated resources [#5716]

  • Refactor: Turn aiida.manage.external.rmq into a package [#5718]

  • Tests: remove legacy tests/utils/configuration.py [#5500]

  • Tests: fix the RPN work chains for the nightly build [#5529]

  • Tests: Manually stop daemon after verdi devel revive test [#5689]

  • Tests: Add verbose info if submit_and_wait times out [#5689]

  • Tests: Do not set default memory for localhost fixture [#5689]

  • Tests: Suppress RabbitMQ and developer version warnings [#5689]

  • Tests: Add the EntryPointManager exposed as entry_points fixture [#5656]

  • Tests: Only reset database connection at end of suite [#5641]

  • Tests: Suppress logging and warnings from temporary profile fixture [#5702]

v2.0.4 - 2022-09-22#

Full changelog

Fixes#

  • Engine: Fix bug that allowed non-storable inputs to be passed to process [#5532]

  • Engine: Fix bug when caching from process with nested outputs [#5538]

  • Archive: Fix bug in archive creation after packing of file repository [#5570]

  • QueryBuilder: apply escape \ in like and ilike for a sqlite backend, such as export archives [#5553]

  • QueryBuilder: Fix bug in distinct queries always projecting the first entity, even if not projected explicitly [#5654]

  • CalcJob: fix bug in local_copy_list provenance exclusion [#5648]

  • Repository.copy_tree: omit subdirectories from path when copying [#5648]

  • Docs: Add intersphinx aliases for __all__ imports. Now the shortcut imports can also be used in third-party packages (e.g. aiida.orm.nodes.node.Node as well as aiida.orm.Node) [#5657]

v2.0.3 - 2022-08-09#

Full changelog

Update of the Dockerfile base image (aiidateam/aiida-prerequisites) to version 0.6.0.

v2.0.2 - 2022-07-13#

Full changelog

Fixes#

  • REST API: treat false as False in URL parsing [#5573]

  • REST API: add support for byte streams through a custom JSON encoder [#5576]

v2.0.1 - 2022-04-28#

Full changelog

Dependencies#

  • Fix incompatibility with click>=8.1 and require click==8.1 as a minimum by @sphuber in [#5504]

v2.0.0 - 2022-04-27#

Full changelog

This release finalises the v2.0.0b1 changes.

Node namespace restructuring ♻️#

Note

The restructuring is fully back-compatible, and existing methods/attributes will continue to work, until aiida-core v3.0.

Deprecations warnings are also currently turned off by default. To identify these deprecations in your code base (for example when running unit tests), activate the AIIDA_WARN_v3 environmental variable:

export AIIDA_WARN_v3=1

The Node class (and thus its subclasses) has many methods and attributes in its public namespace. This has been noted as being a problem for those using auto-completion, since it makes it difficult to select suitable methods and attributes.

These methods/attributes have now been partitioned into “sub-namespaces” for specific purposes:

Node.base.attributes

Interface to the attributes of a node instance.

Node.base.caching

Interface to control caching of a node instance.

Node.base.comments

Interface for comments of a node instance.

Node.base.extras

Interface to the extras of a node instance.

Node.base.links

Interface for links of a node instance.

Node.base.repository

Interface to the file repository of a node instance.

Full list of re-naming

Current name

New name

Collection

Deprecated, use NodeCollection directly

add_comment

Node.base.comments.add

add_incoming

Node.base.links.add_incoming

attributes

Node.base.attributes.all

attributes_items

Node.base.attributes.items

attributes_keys

Node.base.attributes.keys

check_mutability

Node._check_mutability_attributes

clear_attributes

Node.base.attributes.clear

clear_extras

Node.base.extras.clear

clear_hash

Node.base.caching.clear_hash

copy_tree

Node.base.repository.copy_tree

delete_attribute

Node.base.attributes.delete

delete_attribute_many

Node.base.attributes.delete_many

delete_extra

Node.base.extras.delete

delete_extra_many

Node.base.extras.delete_many

delete_object

Node.base.repository.delete_object

erase

Node.base.repository.erase

extras

Node.base.extras.all

extras_items

Node.base.extras.items

extras_keys

Node.base.extras.keys

get

Deprecated, use Node.objects.get

get_all_same_nodes

Node.base.caching.get_all_same_nodes

get_attribute

Node.base.attributes.get

get_attribute_many

Node.base.attributes.get_many

get_cache_source

Node.base.caching.get_cache_source

get_comment

Node.base.comments.get

get_comments

Node.base.comments.all

get_extra

Node.base.extras.get

get_extra_many

Node.base.extras.get_many

get_hash

Node.base.caching.get_hash

get_incoming

Node.base.links.get_incoming

get_object

Node.base.repository.get_object

get_object_content

Node.base.repository.get_object_content

get_outgoing

Node.base.links.get_outgoing

get_stored_link_triples

Node.base.links.get_stored_link_triples

glob

Node.base.repository.glob

has_cached_links

Node.base.caching.has_cached_links

id

Deprecated, use pk

is_created_from_cache

Node.base.caching.is_created_from_cache

is_valid_cache

Node.base.caching.is_valid_cache

list_object_names

Node.base.repository.list_object_names

list_objects

Node.base.repository.list_objects

objects

collection

open

Node.base.repository.open

put_object_from_file

Node.base.repository.put_object_from_file

put_object_from_filelike

Node.base.repository.put_object_from_filelike

put_object_from_tree

Node.base.repository.put_object_from_tree

rehash

Node.base.caching.rehash

remove_comment

Node.base.comments.remove

repository_metadata

Node.base.repository.metadata

repository_serialize

Node.base.repository.serialize

reset_attributes

Node.base.attributes.reset

reset_extras

Node.base.extras.reset

set_attribute

Node.base.attributes.set

set_attribute_many

Node.base.attributes.set_many

set_extra

Node.base.extras.set

set_extra_many

Node.base.extras.set_many

update_comment

Node.base.comments.update

validate_incoming

Node.base.links.validate_incoming

validate_outgoing

Node.base.links.validate_outgoing

validate_storability

Node._validate_storability

verify_are_parents_stored

Node._verify_are_parents_stored

walk

Node.base.repository.walk

IPython integration improvements 👌#

The aiida IPython magic commands are now available to load via:

%load_ext aiida

As well as the previous %aiida magic command, to load a profile, one can also use the %verdi magic command. This command runs the verdi CLI using the currently loaded profile of the IPython/Jupyter session.

%verdi status

See the Basic Tutorial for example usage.

New SqliteTempBackend ✨#

The SqliteTempBackend utilises an in-memory SQLite database to store data, allowing it to be transiently created/destroyed within a single Python session, without the need for Postgresql.

As such, it is useful for demonstrations and testing purposes, whereby no persistent storage is required.

To load a temporary profile, you can use the following code:

from aiida import load_profile
from aiida.storage.sqlite_temp import SqliteTempBackend

profile = load_profile(
    SqliteTempBackend.create_profile(
        'myprofile',
        options={
            'runner.poll.interval': 1
        },
        debug=False
    ),
)

See the Basic Tutorial for example usage.

Key Pull Requests#

Below is a list of some key pull requests that have been merged into version 2.0.0:

  • Node namespace re-structuring:

    • 🔧 MAINTAIN: Add warn_deprecation function, Node.base, and move NodeRepositoryMixin -> NodeRepository by @chrisjsewell in #5472

    • ♻️ REFACTOR: EntityAttributesMixin -> NodeAttributes by @chrisjsewell in #5442

    • ♻️ REFACTOR: Move methods to Node.comments by @chrisjsewell in #5446

    • ♻️ REFACTOR: EntityExtrasMixin -> EntityExtras by @chrisjsewell in #5445

    • ♻️ REFACTOR: Move link related methods to Node.base.links by @sphuber in #5480

    • ♻️ REFACTOR: Move caching related methods to Node.base.caching by @sphuber in #5483

  • Storage:

    • ✨ NEW: Add SqliteTempBackend by @chrisjsewell in #5448

    • 👌 IMPROVE: Move default user caching to StorageBackend by @chrisjsewell in #5460

    • 👌 IMPROVE: Add JSON filtering for SQLite backends by @chrisjsewell in #5448

  • ORM:

    • 👌 IMPROVE: StructureData: allow to be initialised without a specified cell by @ltalirz in #5341

  • Processing:

    • 👌 IMPROVE: Allow engine.run to work without RabbitMQ by @chrisjsewell in #5448

    • 👌 IMPROVE: JobTemplate: change CodeInfo to JobTemplateCodeInfo in codes_info by @unkcpz in #5350

      • This is required for a containerized code implementation

    • 👌 IMPROVE: Add option to use double quotes for Code and Computer CLI arguments by @unkcpz in #5478

  • Transport and Scheduler:

    • 👌 IMPROVE: SlurmScheduler: Parse out-of-walltime and out-of-memory errors from stderr by @sphuber in #5458

    • 👌 IMPROVE: CalcJob: always call Scheduler.parse_output by @sphuber in #5458

    • 👌 IMPROVE: Computer: fallback on transport for get_minimum_job_poll_interval default by @sphuber in #5457

  • IPython:

    • ✨ NEW: Add %verdi IPython magic by @chrisjsewell in #5448

  • Dependencies:

    • ♻️ REFACTOR: drop the python-dateutil library by @sphuber

v2.0.0b1 - 2022-03-15#

Full changelog

The version 2 release of aiida-core largely focusses on major improvements to the design of data storage within AiiDA, as well as updates to core dependencies and removal of deprecated APIs.

Assuming users have already addressed deprecation warnings from aiida-core v1.6.x, there should be limited impact on existing code. For plugin developers, the AiiDA 2.0 plugin migration guide provides a step-by-step guide on how to update their plugins.

For existing profiles and archives, a migration will be required, before they are compatible with the new version.

Tip

Before updating your aiida-core installation, it is advisable to make sure you create a full backup of your profiles, using the current version of aiida-core you have installed. For backup instructions, using aiida-core v1.6.7, see this documentation.

Python support updated to 3.8 - 3.10 ⬆️#

Following the NEP 029 timeline, support for Python 3.7 is dropped as of December 26 2021, and support for Python 3.10 is added.

Plugin entry point updates 🧩#

AiiDA’s use of entry points, to allow plugins to extend the functionality of AiiDA, is described in the plugins topic section.

The use of reentry scan, for loading plugin entry points, is no longer necessary.

Use of the reentry dependency has been replaced by the built-in importlib.metadata library. This library requires no additional loading step.

All entry points provided by aiida-core now start with a core. prefix, to make their origin more explicit and respect the naming guidelines of entry points in the AiiDA ecosystem. The old names are still supported so as to not suddenly break existing code based on them, but they have now been deprecated. For example:

from aiida.plugins import DataFactory
Int = DataFactory('int')  # Old name
Int = DataFactory('core.int')  # New name

Note that entry point names are also used on the command line. For example:

$ verdi computer setup -L localhost -T local -S direct
# now changed to
$ verdi computer setup -L localhost -T local -S core.direct

Improvements to the AiiDA storage architecture ♻️#

Full details on the AiiDA storage architecture are available in the storage architecture section.

The storage refactor incorporates four major changes:

  • The django and sqlalchemy storage backends have been merged into a single psql_dos backend (PostgreSQL + Disk-Objectstore).

  • The file system node repository has been replaced with an object store implementation.

    • The object store automatically deduplicates files, and allows for the compression of many objects into a single file, thus significantly reducing the number of files on the file system and memory utilisation (by orders of magnitude).

    • Note, to make full use of object compression, one should periodically run verdi storage maintain.

    • See the repository design section for details.

  • Command-line interaction with a profile’s storage has been moved from verdi database to verdi storage.

  • The AiiDA archive format has been redesigned as the sqlite_zip storage backend.

    • See the sqlite_zip storage format for details.

    • The new format allows for streaming of data during exports and imports, significantly reducing both the time and memory utilisation of these actions.

    • The archive can now be loaded directly as a (read-only) profile, without the need to import it first, see this Jupyter Notebook tutorial.

The storage redesign also allows for profile switching, within the same Python process, and profile access within a context manager. For example:

from aiida import load_profile, profile_context, orm

with profile_context('my_profile_1'):
    # The profile will be loaded within the context
    node_from_profile_1 = orm.load_node(1)
    # then the profile will be unloaded automatically

# load a global profile
load_profile('my_profile_2')
node_from_profile_2 = orm.load_node(1)

# switch to a different global profile
load_profile('my_profile_3', allow_switch=True)
node_from_profile_3 = orm.load_node(1)

See How to interact with AiiDA for more details.

On first using aiida-core v2.0, your AiiDA configuration will be automatically migrated to the new version (this can be reverted by verdi config downgrade). To update existing profiles and archives to the new storage formats, simply use verdi storage migrate and verdi archive migrate, respectively.

Important

The migration of large storage repositories is a potentially time-consuming process. It may take several hours to complete, depending on the size of the repository. It is also advisable to make a full manual backup of any AiiDA setup with important data: see the installation management section for more information.

See also this testing of profile migrations, for some indicative timings.

Improvements to the AiiDA ORM 👌#

Node repository#

Inline with the storage improvements, Node methods associated with the repository have some backwards incompatible changes:

Node repository method changes

Altered:

  • FileType: moved from aiida.orm.utils.repository to aiida.repository.common

  • File: moved from aiida.orm.utils.repository to aiida.repository.common

  • File: changed from namedtuple to class

  • File: can no longer be iterated over

  • File: type attribute was renamed to file_type

  • Node.put_object_from_tree: path argument was renamed to filepath

  • Node.put_object_from_file: path argument was renamed to filepath

  • Node.put_object_from_tree: key argument was renamed to path

  • Node.put_object_from_file: key argument was renamed to path

  • Node.put_object_from_filelike: key argument was renamed to path

  • Node.get_object: key argument was renamed to path

  • Node.get_object_content: key argument was renamed to path

  • Node.open: key argument was renamed to path

  • Node.list_objects: key argument was renamed to path

  • Node.list_object_names: key argument was renamed to path

  • SinglefileData.open: key argument was renamed to path

  • Node.open: can no longer be called without context manager

  • Node.open: only mode r and rb are supported, use put_object_from_ methods instead

  • Node.get_object_content: only mode r and rb are supported

  • Node.put_object_from_tree: argument contents_only was removed

  • Node.put_object_from_tree: argument force was removed

  • Node.put_object_from_file: argument force was removed

  • Node.put_object_from_filelike: argument force was removed

  • Node.delete_object: argument force was removed

Added:

  • Node.walk

  • Node.copy_tree

  • Node.is_valid_cache setter

  • Node.objects.iter_repo_keys

Additionally, Node.open should always be used as a context manager, for example:

with node.open('filename.txt') as handle:
    content = handle.read()

QueryBuilder#

When using the QueryBuilder to query the database, the following changes have been made:

  • The Computer’s name field is now replaced with label (as previously deprecated in v1.6)

  • The QueryBuilder.queryhelp attribute is deprecated, for the as_dict (and from_dict) methods

  • The QueryBuilder.first method now allows the flat argument, which will return a single item, instead of a list of one item, if only a single projection is defined.

For example:

from aiida.orm import QueryBuilder, Computer
query = QueryBuilder().append(Computer, filters={'label': 'localhost'}, project=['label']).as_dict()
QueryBuilder.from_dict(query).first(flat=True)  # -> 'localhost'

For further information, see How to find and query for data.

Dict usage#

The Dict class has been updated to support more native dict behaviour:

  • Initialisation can now use Dict({'a': 1}), instead of Dict(dict={'a': 1}). This is also the case for List([1, 2]).

  • Equality (==/!=) comparisons now compare the dictionaries, rather than the UUIDs

  • The contains (in) operator now returns True if the dictionary contains the key

  • The items method iterates a list of (key, value) pairs

For example:

from aiida.orm import Dict

d1 = Dict({'a': 1})
d2 = Dict({'a': 1})

assert d1.uuid != d2.uuid
assert d1 == d2
assert not d1 != d2

assert 'a' in d1

assert list(d1.items()) == [('a', 1)]

New data types#

Two new built-in data types have been added:

EnumData

A data plugin that wraps a Python enum.Enum instance.

JsonableData

A data plugin that allows one to easily wrap existing objects that are JSON-able (via an as_dict method).

See the data types section for more information.

Improvements to the AiiDA process engine 👌#

CalcJob API#

A number of minor improvements have been made to the CalcJob API:

  • Both numpy arrays and Enum instances can now be serialized on process checkpoints.

  • The Calcjob.spec.metadata.options.rerunnable option allows to specify whether the calculation can be rerun or requeued (dependent on the scheduler). Note, this should only be applied for idempotent codes.

  • The Calcjob.spec.metadata.options.environment_variables_double_quotes option allows for double-quoting of environment variable declarations. In particular, this allows for use of the $ character in the environment variable name, e.g. export MY_FILE="$HOME/path/my_file".

  • CalcJob.local_copy_list now allows for specifying entire directories to be copied to the local computer, in addition to individual files. Note that the directory itself won’t be copied, just its contents.

  • WorkChain.to_context now allows . delimited namespacing, which generate nested dictionaries. See Nested context keys for more information.

Importing existing computations#

The new CalcJobImporter class has been added, to define importers for computations completed outside of AiiDA. These can help onboard new users to your AiiDA plugin. For more information, see Writing importers for existing computations.

Scheduler plugins#

Plugin’s implementation of Scheduler._get_submit_script_header should now utilise Scheduler._get_submit_script_environment_variables, to format environment variable declarations, rather than handling it themselves. See the exemplar changes in #5283.

The Scheduler.get_valid_transports() method has also been removed, use get_entry_point_names('aiida.schedulers') instead (see get_entry_point_names()).

See Scheduler plugins for more information.

Transport plugins#

The SshTransport now supports the SSH ProxyJump option, for tunnelling through other SSH hosts. See How to setup SSH connections for more information.

Transport plugins now support also transferring bytes (rather than only Unicode strings) in the stdout/stderr of “remote” commands (see #3787). The required changes for transport plugins:

  • rename the exec_command_wait function in your plugin implementation with exec_command_wait_bytes

  • ensure the method signature follows exec_command_wait_bytes(), and that stdin accepts a bytes object.

  • return bytes for stdout and stderr (most probably internally you are already getting bytes - just do not decode them to strings)

For an exemplar implementation, see exec_command_wait_bytes(), or see Transport plugins for more information.

The Transport.get_valid_transports() method has also been removed, use get_entry_point_names('aiida.transports') instead (see get_entry_point_names()).

Improvements to the AiiDA command-line 👌#

The AiiDA command-line interface (CLI) can now be accessed as both verdi and /path/to/bin/python -m aiida.

The underlying dependency for this CLI, click, has been updated to version 8, which contains built-in tab-completion support, to replace the old click-completion. The completion works the same, except that the string that should be put in the activation script to enable it is now shell-dependent. See Activating tab-completion for more information.

Logging for the CLI has been updated, to standardise its use across all CLI commands. This means that all commands include the option:

  -v, --verbosity [notset|debug|info|report|warning|error|critical]
                                  Set the verbosity of the output.

By default the verbosity is set to REPORT (see verdi config list), which relates to using Logger.report, as defined in report().

The following specific changes and improvements have been made to the CLI commands:

verdi storage (replaces verdi database)

This command group replaces the verdi database command group, which is now deprecated, in order to represent its interaction with the full profile storage (not just database).

verdi storage info provides information about the entities contained for a profile.

verdi storage maintain has also been added, to allow for maintenance of the storage, for example, to optimise the storage size.

verdi archive version and verdi archive info (replace verdi archive inspect)

This change synchronises the commands with the new verdi storage version and verdi storage info commands.

verdi group move-nodes

This command moves nodes from a source group to a target group (removing them from one and adding them to the other).

verdi code setup

There is a small change to the order of prompts, in interactive mode.

The uniqueness of labels is now validated, for both remote and local codes.

verdi code test

Run tests for a given code to check whether it is usable, including whether remote executable files are available.

See AiiDA Command Line for more information.

Development improvements#

The build tool for aiida-core has been changed from setuptools to flit. This allows for the project metadata to be fully specified in the pyproject.toml file, using the PEP 621 format. Note, editable installs (using the -e flag for pip install) of aiida-core now require pip>=21.

Type annotations have been added to most of the code base. Plugin developers can use mypy to check their code against the new type annotations.

All module level imports are now defined explicitly in __all__. See Overview of public API for more information.

The aiida.common.json module is now deprecated. Use the json standard library instead.

Changes to the plugin test fixtures 🧪#

The deprecated AiidaTestCase class has been removed, in favour of the AiiDA pytest fixtures, which can be loaded in your conftest.py using:

pytest_plugins = ['aiida.manage.tests.pytest_fixtures']

The fixtures clear_database, clear_database_after_test, clear_database_before_test are now deprecated, in favour of the aiida_profile_clean fixture, which ensures (before the test) the default profile is reset with clean storage, and that all previous resources are closed If you only require the profile to be reset before a class of tests, then you can use aiida_profile_clean_class.

Key Pull Requests#

Below is a list of some key pull requests that have been merged into version 2.0.0b1:

  • Storage and migrations:

    • ♻️ REFACTOR: Implement the new file repository by @sphuber in #4345

    • ♻️ REFACTOR: New archive format by @chrisjsewell in #5145

    • ♻️ REFACTOR: Remove QueryManager by @chrisjsewell in #5101

    • ♻️ REFACTOR: Fully abstract QueryBuilder by @chrisjsewell in #5093

    • ✨ NEW: Add Backend bulk methods by @chrisjsewell in #5171

    • ⬆️ UPDATE: SQLAlchemy v1.4 (v2 API) by @chrisjsewell in #5103 and #5122

    • 👌 IMPROVE: Configuration migrations by @chrisjsewell in #5319

    • ♻️ REFACTOR: Remove Django storage backend by @chrisjsewell in #5330

    • ♻️ REFACTOR: Move archive backend to aiida/storage by @chrisjsewell in 5375

    • 👌 IMPROVE: Use sqlalchemy.func for JSONB QB filters by @ltalirz in #5393

    • ✨ NEW: Add Mechanism to lock profile access by @ramirezfranciscof in #5270

    • ✨ NEW: Add verdi storage CLI by @ramirezfranciscof in #4965 and #5156

  • ORM API:

    • ♻️ REFACTOR: Add the core. prefix to all entry points by @sphuber in #5073

    • 👌 IMPROVE: Replace InputValidationError with ValueError and TypeError by @sphuber in #4888

    • 👌 IMPROVE: Add Node.walk method to iterate over repository content by @sphuber in #4935

    • 👌 IMPROVE: Add Node.copy_tree method by @sphuber in #5114

    • 👌 IMPROVE: Add Node.is_valid_cache setter property by @sphuber in #5114

    • 👌 IMPROVE: Add Node.objects.iter_repo_keys by @chrisjsewell in #5114

    • 👌 IMPROVE: Allow storing Decimal in Node.attributes by @dev-zero in #4964

    • 🐛 FIX: Initialising a Node with a User by @chrisjsewell in #5114

    • 🐛 FIX: Deprecate double underscores in LinkManager contains by @sphuber in #5067

    • ♻️ REFACTOR: Rename name field of Computer to label by @sphuber in #4882

    • ♻️ REFACTOR: QueryBuilder.queryhelp -> QueryBuilder.as_dict by @chrisjsewell in #5081

    • 👌 IMPROVE: Add AuthInfo joins to QueryBuilder by @chrisjsewell in #5195

    • 👌 IMPROVE: QueryBuilder.first add flat keyword by @sphuber in #5410

    • 👌 IMPROVE: Add Computer.default_memory_per_machine attribute by @yakutovicha in #5260

    • 👌 IMPROVE: Add Code.validate_remote_exec_path method to check executable by @sphuber in #5184

    • 👌 IMPROVE: Allow source to be passed as a keyword to Data.__init__ by @sphuber in #5163

    • 👌 IMPROVE: Dict.__init__ and List.__init__ by @mbercx in #5165

    • ‼️ BREAKING: Compare Dict nodes by content by @mbercx in #5251

    • 👌 IMPROVE: Implement the Dict.__contains__ method by @sphuber in #5251

    • 👌 IMPROVE: Implement Dict.items() method by @mbercx in #5251

    • 🐛 FIX: BandsData.show_mpl allow NaN values by @PhilippRue in #5024

    • 🐛 FIX: Replace KeyError with AttributeError in TrajectoryData methods by @Crivella in #5015

    • ✨ NEW: EnumData data plugin by @sphuber in #5225

    • ✨ NEW: JsonableData data plugin by @sphuber in #5017

    • 👌 IMPROVE: Register List class with to_aiida_type dispatch by @sphuber in #5142

    • 👌 IMPROVE: Register EnumData class with to_aiida_type dispatch by @sphuber in #5314

  • Processing:

    • ✨ NEW: CalcJob.get_importer() to import existing calculations, run outside of AiiDA by @sphuber in #5086

    • ✨ NEW: ProcessBuilder._repr_pretty_ ipython representation by @mbercx in #4970

    • 👌 IMPROVE: Allow Enum types to be serialized on ProcessNode.checkpoint by @sphuber in #5218

    • 👌 IMPROVE: Allow numpy arrays to be serialized on ProcessNode.checkpoint by @greschd in #4730

    • 👌 IMPROVE: Add Calcjob.spec.metadata.options.rerunnable to requeue/rerun calculations by @greschd in #4707

    • 👌 IMPROVE: Add Calcjob.spec.metadata.options.environment_variables_double_quotes to escape environment variables by @unkcpz in #5349

    • 👌 IMPROVE: Allow directories in CalcJob.local_copy_list by @sphuber in #5115

    • 👌 IMPROVE: Add support for . namespacing in the keys for WorkChain.to_context by @dev-zero in #4871

    • 👌 IMPROVE: Handle namespaced outputs in BaseRestartWorkChain by @unkcpz in #4961

    • 🐛 FIX: Nested namespaces in ProcessBuilderNamespace by @sphuber in #4983

    • 🐛 FIX: Ensure ProcessBuilder instances do not interfere by @sphuber in #4984

    • 🐛 FIX: Raise when Process.exposed_outputs gets non-existing namespace by @sphuber in #5265

    • 🐛 FIX: Catch AttributeError for unloadable identifier in ProcessNode.is_valid_cache by @sphuber in #5222

    • 🐛 FIX: Handle CalcInfo.codes_run_mode when CalcInfo.codes_info contains multiple codes by @unkcpz in #4990

    • 🐛 FIX: Check for recycled circus PID by @dev-zero in #5086

  • Scheduler/Transport:

    • 👌 IMPROVE: Specify abstract methods on Transport by @chrisjsewell in #5242

    • ✨ NEW: Add support for SSH proxy_jump by @dev-zero in #4951

    • 🐛 FIX: Daemon hang when passing None as job_id by @ramirezfranciscof in #4967

    • 🐛 FIX: Avoid deadlocks when retrieving stdout/stderr via SSH by @giovannipizzi in #3787

    • 🐛 FIX: Use sanitised variable name in SGE scheduler job title by @mjclarke94 in #4994

    • 🐛 FIX: listdir method with pattern for SSH by @giovannipizzi in #5252

    • 👌 IMPROVE: DirectScheduler: use num_cores_per_mpiproc if defined in resources by @sphuber in #5126

    • 👌 IMPROVE: Add abstract generation of submit script env variables to Scheduler by @sphuber in #5283

  • CLI:

    • ✨ NEW: Allow for CLI usage via python -m aiida by @chrisjsewell in #5356

    • ⬆️ UPDATE: click==8.0 and remove click-completion by @sphuber in #5111

    • ♻️ REFACTOR: Replace verdi database commands with verdi storage by @ramirezfranciscof in #5228

    • ✨ NEW: Add verbosity control by @sphuber in #5085

    • ♻️ REFACTOR: Logging verbosity implementation by @sphuber in #5119

    • ✨ NEW: Add verdi group move-nodes command by @mbercx in #4428

    • 👌 IMPROVE: verdi code setup: validate the uniqueness of label for local codes by @sphuber in #5215

    • 👌 IMPROVE: GroupParamType: store group if created by @sphuber in #5411

    • 👌 IMPROVE: Show #procs/machine in verdi computer show by @dev-zero in #4945

    • 👌 IMPROVE: Notify users of runner usage in verdi process list by @ltalirz in #4663

    • 👌 IMPROVE: Set localhost as default for database hostname in verdi setup by @sphuber in #4908

    • 👌 IMPROVE: Make verdi group messages consistent by @CasperWA in #4999

    • 🐛 FIX: verdi calcjob cleanworkdir command by @zhubonan in #5209

    • 🔧 MAINTAIN: Add verdi devel run-sql by @chrisjsewell in #5094

  • REST API:

    • ⬆️ UPDATE: Update to flask~=2.0 for rest extra by @sphuber in #5366

    • 👌 IMPROVE: Error message when flask not installed by @ltalirz in #5398

    • 👌 IMPROVE: Allow serving of contents of ArrayData by @JPchico in #5425

    • 🐛 FIX: REST API date-time query by @NinadBhat in #4959

  • Developers:

    • 🔧 MAINTAIN: Move to flit for PEP 621 compliant package build by @chrisjsewell in #5312

    • 🔧 MAINTAIN: Make __all__ imports explicit by @chrisjsewell in #5061

    • 🔧 MAINTAIN: Add pre-commit.ci by @chrisjsewell in #5062

    • 🔧 MAINTAIN: Add isort pre-commit hook by @chrisjsewell in #5151

    • ⬆️ UPDATE: Drop support for Python 3.7 by @sphuber in #5307

    • ⬆️ UPDATE: Support Python 3.10 by @csadorf in #5188

    • ♻️ REFACTOR: Remove reentry requirement by @chrisjsewell in #5058

    • ♻️ REFACTOR: Remove simplejson by @sphuber in #5391

    • ♻️ REFACTOR: Remove ete3 dependency by @ltalirz in #4956

    • 👌 IMPROVE: Replace deprecated imp with importlib by @DirectriX01 in #4848

    • ⬆️ UPDATE: sphinx~=4.1 (+ sphinx extensions) by @chrisjsewell in #5420

    • 🧪 CI: Move time consuming tests to separate nightly workflow by @sphuber in #5354

    • 🧪 TESTS: Entirely remove AiidaTestCase by @chrisjsewell in #5372

Contributors 🎉#

Thanks to all contributors: Contributor Graphs

Including first-time contributors:

  • @DirectriX01 made their first contribution in [#4848]

  • @mjclarke94 made their first contribution in [#4994]

  • @janssenhenning made their first contribution in [#5064]

v1.6.7 - 2022-03-07#

full changelog

The markupsafe dependency specification was moved to install_requires

v1.6.6 - 2022-03-07#

full changelog

Bug fixes 🐛#

  • DirectScheduler: remove the -e option for bash invocation [#5264]

  • Replace deprecated matplotlib config option ‘text.latex.preview’ [#5233]

Dependencies#

  • Add upper limit markupsafe<2.1 to fix the documentation build [#5371]

  • Add upper limit pytest-asyncio<0.17 [#5309]

Devops 🔧#

  • CI: move Jenkins workflow to nightly GHA workflow [#5277]

  • Docs: replace CircleCI build with ReadTheDocs [#5279]

  • CI: run certain workflows only on main repo, not on forks [#5091]

  • Revise Docker image build [#4997]

v1.6.5 - 2021-08-13#

full changelog

This patch release contains a number of helpful bug fixes and improvements.

Improvements 👌#

  • Add support for the ProxyJump SSH config option for seting up an arbitrary number of proxy jumps without additional processes by creating TCP channels over existing SSH connections. This provides improved control over the lifetime of the different connections. See SSH configuration for further details. [#4951]

  • Allow numpy arrays to be serialized to a process checkpoint. [#4730)])

  • Add the _merge method to ProcessBuilder, to update the builder with a nested dictionary. [#4983)])

  • verdi setup: Set the defaut database hostname as localhost. [#4908]

  • Allow Node.__init__ to be constructed with a specific User node. [#4977]

  • Minimize database logs of failed schema version retrievals. [#5056]

  • Remove duplicate call of normal callback for InteractiveOption. [#5064]

  • Update requirement pyyaml~=5.4, which contains critical security fixes. [#5060]

Bug Fixes 🐛#

  • Fix regression issue with __contains__ operator in LinkManager, when using double underscores, e.g. for 'some__nested__namespace' in calc.inputs. #5067

  • Stop deprecation warning being shown when tab-completing incoming and outgoing node links. [#5011]

  • Stop possible command hints being shown when attempting to tab complete verdi commands that do not exist. [#5012]

  • Do not use get_detailed_job_info when retrieving a calculation job, if no job id is set. [#4967]

  • Race condition when two processes try to create the same Folder/SandboxFolder, [#4912]

  • Return the whole nested namespace when using BaseRestartWorkChain.result. [#4961]

  • Use numpy.nanmin and numpy.nanmax for computing y-limits of BandsData matplotlib methods. [#5024]

  • Use sanitized job title with SgeScheduler scheduler. [#4994]

v1.6.4 - 2021-06-23#

full changelog

This is a patch release to pin psycopg2-binary to version 2.8.x, to avoid an issue with database creation in version 2.9 (#4989).

v1.6.3 - 2021-04-28#

full changelog | GitHub contributors page for this release

This is a patch release to fix a bug that was introduced in v1.6.2 that would cause a number of verdi commands to fail due to a bug in the with_dbenv decorator utility.

Bug fixes#

  • Fix aiida.cmdline.utils.decorators.load_backend_if_not_loaded [#4878]

v1.6.2 - 2021-04-28#

full changelog | GitHub contributors page for this release

Bug fixes#

  • CLI: Use the proper proxy command for verdi calcjob gotocomputer if configured as such [#4761]

  • Respect nested output namespaces in Process.exposed_outputs [#4863]

  • NodeLinkManager now properly regenerates original nested namespaces from the flat link labels stored in the database. This means one can now do node.outputs.some.nested.output instead of having to do node.outputs.some__nested__output. The same goes for node.inputs [#4625]

  • Fix aiida.cmdline.utils.decorators.with_dbenv always loading the database. Now it will only load the database if not already loaded, as intended [#4865]

Features#

  • Add the account option to the LsfScheduler scheduler plugin [#4832]

Documentation#

  • Update ssh proxycommand section with instructions on how to handle cases where the SSH key needs to be specified for the proxy server [#4839]

  • Add the “How to extend workflows” section, explaining the use of the expose_inputs and expose_outputs features, as well as nested namespaces [#4562]

  • Add help in intro for when quicksetup fails due to problems autodetecting the PostgreSQL settings [#4838]

v1.6.1 - 2021-03-31#

full changelog | GitHub contributors page for this release

This patch release is primarily intended to fix a regression in the aiida_profile test fixture, used by plugin developers, causing config validation errors (#4831).

Other additions:

  • ✨ NEW: Added structure.data.import entry-point, allowing for plugins to define file-format specific sub-commands of verdi data structure import (#4427).

  • ✨ NEW: Added --label and --group options to verdi data structure import, which apply a label/group to all structures being imported (#4429).

  • ⬆️ UPDATE: psgu dependency increased to v0.2.x. This fixes a bug in verdi quicksetup, when used on the Windows Subsystem for Linux (WSL) platform (#4834).

  • 🐛 FIX: metadata.options.max_memory_kb is now ignored when using the direct scheduler (#4825). This was previously imposing a a virtual memory limit with ulimit -v, which is very different to the physical memory limit that other scheduler plugins impose. No straightforward way exists to directly limit the physical memory usage for this scheduler.

  • 🐛 FIX: Added __str__ method to the Orbital class, fixing a recursion error (#4829).

v1.6.0 - 2021-03-15#

full changelog | GitHub contributors page for this release

As well as introducing a number of improvements and new features listed below, this release marks the “under-the-hood” migration from the tornado package to the Python built-in module asyncio, for handling asynchronous processing within the AiiDA engine. This removes a number of blocking dependency version clashes with other tools, in particular with the newest Jupyter shell and notebook environments. The migration does not present any backward incompatible changes to AiiDA’s public API. A substantial effort has been made to test and debug the new implementation, and ensure it performs at least equivalent to the previous code (or improves it!), but please let us know if you uncover any additional issues.

This release also drops support for Python 3.6 (testing is carried out against 3.7, 3.8 and 3.9).

NOTE: v1.6 is tentatively intended to be the final minor v1.x release before v2.x, that will include a new file repository implementation and remove all deprecated code.

New calculation features ✨#

The additional_retrieve_list metadata option has been added to CalcJob (#4437). This new option allows one to specify additional files to be retrieved on a per-instance basis, in addition to the files that are already defined by the plugin to be retrieved.

A new namespace stash has bee added to the metadata.options input namespace of the CalcJob process (#4424). This option namespace allows a user to specify certain files that are created by the calculation job to be stashed somewhere on the remote. This can be useful if those files need to be stored for a longer time than the scratch space (where the job was run) is available for, but need to be kept on the remote machine and not retrieved. Examples are files that are necessary to restart a calculation but are too big to be retrieved and stored permanently in the local file repository.

See Stashing files on the remote for more details.

The new TransferCalcjob plugin (#4194) allows the user to copy files between a remote machine and the local machine running AiiDA. More specifically, it can do any of the following:

  • Take any number of files from any number of RemoteData folders in a remote machine and copy them in the local repository of a single newly created FolderData node.

  • Take any number of files from any number of FolderData nodes in the local machine and copy them in a single newly created RemoteData folder in a given remote machine.

See the Transferring data how-to for more details.

Profile configuration improvements 👌#

The way the global/profile configuration is accessed has undergone a number of distinct changes (#4712):

  • When loaded, the config.json (found in the .aiida folder) is now validated against a JSON Schema that can be found in aiida/manage/configuration/schema.

  • The schema includes a number of new global/profile options, including: transport.task_retry_initial_interval, transport.task_maximum_attempts, rmq.task_timeout and logging.aiopika_loglevel (#4583).

  • The cache_config.yml has now also been deprecated and merged into the config.json, as part of the profile options. This merge will be handled automatically, upon first load of the config.json using the new AiiDA version.

In-line with these changes, the verdi config command has been refactored into separate commands, including verdi config list, verdi config set, verdi config unset and verdi config caching.

See the Configuring profile options and Configuring caching how-tos for more details.

Command-line additions and improvements 👌#

In addition to verdi config, numerous other new commands and options have been added to verdi:

  • Deprecated verdi export and verdi import commands (replaced by new verdi archive) (#4710)

  • Added verdi group delete --delete-nodes, to also delete the nodes in a group during its removal (#4578).

  • Improved verdi group remove-nodes command to warn when requested nodes are not in the specified group (#4728).

  • Added exception to the projection mapping of verdi process list, for example to use in debugging as: verdi process list -S excepted -P ctime pk exception (#4786).

  • Added verdi database summary (#4737): This prints a summary of the count of each entity and (optionally) the list of unique identifiers for some entities.

  • Improved verdi process play performance, by only querying for active processes with the --all flag (#4671)

  • Added the verdi database version command (#4613): This shows the schema generation and version of the database of the given profile, useful mostly for developers when debugging.

  • Improved verdi node delete performance (#4575): The logic has been re-written to greatly reduce the time to delete large amounts of nodes.

  • Fixed verdi quicksetup --non-interactive, to ensure it does not include any user prompts (#4573)

  • Fixed verdi --version when used in editable mode (#4576)

API additions and improvements 👌#

The base Node class now evaluates equality based on the node’s UUID (#4753). For example, loading the same node twice will always resolve as equivalent: load_node(1) == load_node(1). Note that existing, class specific, equality relationships will still override the base class behaviour, for example: Int(99) == Int(99), even if the nodes have different UUIDs. This behaviour for subclasses is still under discussion at: aiidateam/aiida-core#1917

When hashing nodes for use with the caching features, -0. is now converted to 0., to reduce issues with differing hashes before/after node storage (#4648). Known failure modes for hashing are now also raised with the HashingError exception (#4778).

Both aiida.tools.delete_nodes (#4578) and aiida.orm.to_aiida_type (#4672) have been exposed for use in the public API.

A pathlib.Path instance can now be used for the file argument of SinglefileData (#3614)

Type annotations have been added to all inputs/outputs of functions and methods in aiida.engine (#4669) and aiida/orm/nodes/processes (#4772). As outlined in PEP 484, this improves static code analysis and, for example, allows for better auto-completion and type checking in many code editors.

New REST API Query endpoint ✨#

The /querybuilder endpoint is the first POST method available for AiiDA’s RESTful API (#4337)

The POST endpoint returns what the QueryBuilder would return, when providing it with a proper queryhelp dictionary (see the documentation here). Furthermore, it returns the entities/results in the “standard” REST API format - with the exception of link_type and link_label keys for links (these particular keys are still present as type and label, respectively).

For security, POST methods can be toggled on/off with the verdi restapi --posting/--no-posting options (it is on by default). Although note that this option is not yet strictly public, since its naming may be changed in the future!

See AiiDA REST API documentation for more details.

Additional Changes#

  • Fixed the direct scheduler which, in combination with SshTransport, was hanging on submit command (#4735). In the ssh transport, to emulate ‘chdir’, the current directory is now kept in memory, and every command prepended with cd FOLDER_NAME && ACTUALCOMMAND.

  • In aiida.tools.ipython.ipython_magics, load_ipython_extension has been deprecated in favour of register_ipython_extension (#4548).

  • Refactored .ci/ folder to make tests more portable and easier to understand (#4565) The ci/ folder had become cluttered, containing configuration and scripts for both the GitHub Actions and Jenkins CI. This change moved the GH actions specific scripts to .github/system_tests, and refactored the Jenkins setup/tests to use molecule in the .molecule/ folder.

  • For aiida-core development, the pytest requires_rmq marker and config_with_profile fixture have been added (#4739 and #4764)

v1.5.2 - 2020-12-07#

Note: release v1.5.1 was skipped due to a problem with the uploaded files to PyPI.

Bug fixes#

  • Dict: accessing an inexistent key now raises a KeyError (instead of AttributeError) [#4577]

  • Config: make writing to disk as atomic as possible [#4607]

  • Config: do not overwrite when loaded and not migrated [#4605]

  • SqlAlchemy: fix bug in Group extras migration with revision 0edcdd5a30f0 [#4602]

Developers#

  • SqlAlchemy: improve the alembic migration code [#4602] 4607

  • CI: manually install numpy to prevent incompatible releases [#4615]

v1.5.0 - 2020-11-13#

In this minor version release, support for Python 3.9 is added [#4301], while support for Python 3.5 is dropped [#4386]. This version is compatible with all current Python versions that are not end-of-life:

  • 3.6

  • 3.7

  • 3.8

  • 3.9

Features#

  • Process functions (calcfunction and workfunction) can now be submitted to the daemon just like CalcJobs and WorkChains [#4539]

  • REST API: list endpoints at base URL [#4412]

  • REST API: new full_types_count endpoint that counts the number of nodes for each type of node [#4277]

  • ProcessBuilder: allow unsetting of inputs through attribute deletion [#4419]

  • verdi migrate: make --in-place work across different file systems [#4393]

Improvements#

  • Added remaining original documentation that didn’t make it into the first step of the recent major overhaul of v1.3.0

  • verdi process show: order by ctime and print process label [#4407]

  • LinkManager: fix inaccuracy in exception message for non-existent link [#4388]

  • Add reset method toProgressReporterAbstract [#4522]

  • Improve the deprecation warning for Node.open outside context manager [#4434]

Bug fixes#

  • SlurmScheduler: fix bug in validation of job resources [#4555]

  • Fix ZeroDivisionError in worker slots check [#4513]

  • CalcJob: only attempt to clean up the retrieve temporary folder after parsing if it is present [#4379]

  • Add missing entry point groups to the mapping [#4395]

  • REST API: the process_type can now identify pathological empty-stringed or null entries in the database [#4277]

Developers#

  • verdi group delete: deprecate and ignore the --clear option [#4357]

  • Replace old format string interpolation with f-strings [#4400]

  • CI: move pylint configuration to pyproject.toml [#4411]

  • CI: use -e install for tox + add docker-compose for isolated RabbitMQ [#4375]

  • CI: add coverage patch threshold to prevent false positives [#4413]

  • CI: Allow for mypy type checking of third-party imports [#4553]

Dependencies#

  • Update requirement pytest~=6.0 and use pyproject.toml [#4410]

Archive (import/export) refactor#

  • The refactoring goal was to pave the way for the implementation of a new archive format in v2.0.0 ( aiidateamAEP005)

  • Three abstract+concrete interface classes are defined; writer, reader, migrator, which are independent of theinternal structure of the archive. These classes are used within the export/import code.

  • The code in aiida/tools/importexport has been largely re-written, in particular adding aiida/toolsimportexport/archive, which contains this code for interfacing with an archive, and does not require connectionto an AiiDA profile.

  • The export logic has been re-written; to minimise required queries (faster), and to allow for “streaming” datainto the writer (minimise RAM requirement with new format). It is intended that a similiar PR will be made for the import code.

  • A general progress bar implementation is now available in aiida/common/progress_reporter.py. All correspondingCLI commands now also have --verbosity option.

  • Merged PRs:

    • Refactor export archive (#4448 & #4534)

    • Refactor import archive (#4510)

    • Refactor migrate archive (#4532)

    • Add group extras to archive (#4521)

    • Refactor cmdline progress bar (#4504 & #4522)

  • Updated archive version from 0.9 -> 0.10 (#4561

  • Deprecations: export_zip, export_tar, export_tree, extract_zip, extract_tar and extract_treefunctions. silent key-word in the export function

  • Removed: ZipFolder class

v1.4.4#

This patch is a backport for 2 of the fixes in v1.5.2.

Bug fixes#

  • Dict: accessing an inexistent key now raises a KeyError (instead of an AttributeError) [#4616]

Developers#

  • CI: manually install numpy to prevent incompatible releases [#4615]

v1.4.3#

Bug fixes#

  • RabbitMQ: update topika requirement to fix SSL connections and remove validation of broker_parameters from profile [#4542]

  • Fix UnboundLocalError in aiida.cmdline.utils.edit_multiline_template, which affected verdi code/computer setup [#4436]

v1.4.2#

Critical bug fixes#

  • CalcJob: make sure local_copy_list files do not end up in the node’s repository folder [#4415]

v1.4.1#

Improvements#

  • verdi setup: forward broker defaults to interactive mode [#4405]

Bug fixes#

  • verdi setup: improve validation and help string of broker virtual host [#4408]

  • Implement next and iter for the Node.open deprecation wrapper [#4399]

  • Dependencies: increase minimum version requirement plumpy~=0.15.1 to suppress noisy warning at end of interpreter that ran processes [#4398]

v1.4.0#

Improvements#

  • Add defaults for configure options of the SshTransport plugin [#4223]

  • verdi status: distinguish database schema version incompatible [#4319]

  • SlurmScheduler: implement parse_output to detect OOM and OOW [#3931]

Features#

  • Make the RabbitMQ connection parameters configurable [#4341]

  • Add infrastructure to parse scheduler output for CalcJobs [#3906]

  • Add support for “peer” authentication with PostgreSQL [#4255]

  • Add the --paused flag to verdi process list [#4213]

  • Make the loglevel of the daemonizer configurable [#4276]

  • Transport: add option to not use a login shell for all commands [#4271]

  • Implement skip_orm option for SqlAlchemy Group.remove_nodes [#4214]

  • Dict: allow setting attributes through setitem and AttributeManager [#4351]

  • CalcJob: allow nested target paths for local_copy_list [#4373]

  • verdi export migrate: add --in-place flag to migrate archive in place [#4220]

Bug fixes#

  • verdi: make --prepend-text and --append-text options properly interactive [#4318]

  • verdi computer test: fix failing result in harmless stderr responses [#4316]

  • QueryBuilder: Accept empty string for entity_type in append method [#4299]

  • verdi status: do not except when no profile is configured [#4253]

  • ArithmeticAddParser: attach output before checking for negative value [#4267]

  • CalcJob: fix bug in retrieve_list affecting entries without wildcards [#4275]

  • TemplateReplacerCalculation: make files namespace dynamic [#4348]

Developers#

  • Rename folder test.fixtures to test.static [#4219]

  • Remove all files from the pre-commit exclude list [#4196]

  • ORM: move attributes/extras methods of frontend and backend nodes to mixins [#4376]

Dependencies#

  • Dependencies: update minimum requirement paramiko~=2.7 [#4222]

  • Depedencies: remove upper limit and allow numpy~=1.17 [#4378]

Deprecations#

  • Deprecate getter and setter methods of Computer properties [#4252]

  • Deprecate methods that refer to a computer’s label as name [#4309]

Changes#

  • BaseRestartWorkChain: do not run process_handler when exit_codes=[] [#4380]

  • SlurmScheduler: always raise for non-zero exit code [#4332]

  • Remove superfluous ERROR_NO_RETRIEVED_FOLDER from CalcJob subclasses [#3906]

v1.3.1#

Bug fixes:#

  • Fix a file handle leak due to the Runner not closing the event loop if it created it itself [#4307]

  • ArithmeticAddParser: attach output before checking for negative value [#4267]

v1.3.0#

Improvements#

  • Comprehensive restructuring and revamp of the online documentation [#4141]

  • Improve defaults for verdi computer configure ssh [#4055]

  • Provenance graphs: enable highlighting specific node classes (and highlight root node by default) [#4081]

Performance#

  • Enable event-based monitoring of work chain child processes (they were being polled every second) [#4154]

  • Increase the default for runner.poll.interval config option from 1 to 60 seconds [#4150]

  • Increase the efficiency of the SqlaGroup.nodes iterator [#4094]

Features#

  • Add a progress bar for export and import related functionality [#3599]

  • Enable loading config.yml files from URL in verdi commands with --config option [#3977]

  • QueryBuilder: add the flat argument to the .all() method [#3945]

  • verdi status: add --no-rmq flag to skip the RabbitMQ check [#4181]

  • Add support for process functions in verdi plugin list [#4117]

  • Allow profile selection in ipython magic %aiida [#4071]

  • Support more complex formula formats in aiida.orm.data.cif.parse_formula [#3954]

Bug fixes#

  • BaseRestartWorkChain: do not assume metadata exists in inputs in run_process [#4210]

  • BaseRestartWorkChain: fix bug in inspect_process [#4166]

  • BaseRestartWorkChain: fix the “unhandled failure mechanism” for dealing with failures of subprocesses [#4155]

  • Fix exception handling in commands calling list_repository_contents [#3968]

  • Fix bug in Code.get_full_text_info [#4083]

  • Fix bug in verdi daemon restart --reset [#3969]

  • Fix tab-completion for LinkManager and AttributeManager [#3985]

  • CalcJobResultManager: fix bug that broke tab completion [#4187]

  • SshTransport.gettree: allow non-existing nested target directories [#4175]

  • CalcJob: move job resource validation to the Scheduler class fixing a problem for the SGE and LSF scheduler plugins [#4192]

  • WorkChain: guarantee to maintain order of appended awaitables [#4156]

  • Add support for binary files to the various verdi cat commands [#4077]

  • Ensure verdi group show --limit respects limit even in raw mode [#4092]

  • QueryBuilder: fix type string filter generation for Group subclasses [#4144]

  • Raise when calling Node.objects.delete for node with incoming links [#4168]

  • Properly handle multiple requests to threaded REST API [#3974]

  • NodeTranslator: do not assume get_export_formats exists [#4188]

  • Only color directories in verdi node repo ls --color [#4195]

Developers#

  • Add arithmetic workflows and restructure calculation plugins [#4124]

  • Add minimal mypy run to the pre-commit hooks. [#4176]

  • Fix timeout in tests.cmdline.commands.test_process:test_pause_play_kill [#4052]

  • Revise update-dependency flow to resolve issue #3930 [#3957]

  • Add GitHub action for transifex upload [#3958]

Deprecations#

  • The get_valid_schedulers class method of the Scheduler class has been deprecated in favor of aiida.plugins.entry_point.get_entry_point_names [#4192]

v1.2.1#

In the fixing of three bugs, three minor features have been added along the way.

Features#

  • Add config option daemon.worker_process_slots to configure the maximum number of concurrent tasks each daemon worker can handle [#3949]

  • Add config option daemon.default_workers to set the default number of workers to be started by verdi daemon start [#3949]

  • CalcJob: make submit script filename configurable through the metadata.options [#3948]

Bug fixes#

  • CalcJob: fix bug in idempotency check of upload transport task [#3948]

  • REST API: reintroduce CORS headers, the lack of which was breaking the Materials Cloud provenance explorer [#3951]

  • Remove the equality operator of ExitCode which caused the serialization of workchains to fail if put in the workchain context [#3940]

Deprecations#

  • The hookup argument of aiida.restapi.run_api and the --hookup option of verdi restapi are deprecated [#3951]

v1.2.0#

Features#

  • ExitCode: make the exit message parameterizable through templates [#3824]

  • GroupPath: a utility to work with virtual Group hierarchies [#3613]

  • Make Group sub classable through entry points [#3882][#3903][#3926]

  • Add auto-complete support for CodeParamType and GroupParamType [#3926]

  • Add export archive migration for Group type strings [#3912]

  • Add the -v/--version option to verdi export migrate [#3910]

  • Add the -l/--limit option to verdi group show [#3857]

  • Add the --order-by/--order-direction options to verdi group list [#3858]

  • Add prepend_text and append_text to aiida_local_code_factory pytest fixture [#3831]

  • REST API: make it easier to call run_api in wsgi scripts [#3875]

  • Plot bands with only one kpoint [#3798]

Bug fixes#

  • Improved validation for CLI parameters [#3894]

  • Ensure unicity when creating instances of Autogroup [#3650]

  • Prevent nodes without registered entry points from being stored [#3886]

  • Fix the RotatingFileHandler configuration of the daemon logger[#3891]

  • Ensure log messages are not duplicated in daemon log file [#3890]

  • Convert argument to str in aiida.common.escaping.escape_for_bash [#3873]

  • Remove the return statement of RemoteData.getfile() [#3742]

  • Support for BandsData nodes without StructureData ancestors [#3817]

Deprecations#

  • Deprecate --group-type option in favor of --type-string for verdi group list [#3926]

Documentation#

  • Docs: link to documentation of other libraries via intersphinx mapping [#3876]

  • Docs: remove extra advanced_plotting from install instructions [#3860]

  • Docs: consistent use of “plugin” vs “plugin package” terminology [#3799]

Developers#

  • Deduplicate code for tests of archive migration code [#3924]

  • CI: use GitHub Actions services for PostgreSQL and RabbitMQ [#3901]

  • Move aiida.manage.external.pgsu to external package pgsu [#3892]

  • Cleanup the top-level directory of the repository [#3738]

  • Remove unused orm.implementation.utils module [#3877]

  • Revise dependency management workflow [#3771]

  • Re-add support for Coverage reports through codecov.io [#3618]

v1.1.1#

Changes#

  • Emit a warning when input port specifies a node instance as default [#3466]

  • BaseRestartWorkChain: require process handlers to be instance methods [#3782]

  • BaseRestartWorkChain: add method to enable/disable process handlers [#3786]

  • Docker container: remove conda activation from configure-aiida.sh script [#3791]

  • Add fixtures to clear the database before or after tests [#3783]

  • verdi status: add the configuration directory path to the output [#3587]

  • QueryBuilder: add support for datetime.date objects in filters [#3796]

Bug fixes#

  • Fix bugs in Node._store_from_cache and Node.repository.erase that could result in calculations not being reused [#3777]

  • Caching: fix configuration spec and validation [#3785]

  • Write migrated config to disk in Config.from_file [#3797]

  • Validate label string at code setup stage [#3793]

  • Reuse prepend_text and append_text in verdi computer/code duplicate [#3788]

  • Fix broken imports of urllib in various locations including verdi import [#3767]

  • Match headers with actual output for verdi data structure list [#3756]

  • Disable caching for the Data node subclass (this should not affect usual caching behavior) [#3807]

v1.1.0#

Nota Bene: although this is a minor version release, the support for python 2 is dropped (#3566) following the reasoning outlined in the corresponding AEP001. Critical bug fixes for python 2 will be supported until July 1 2020 on the v1.0.* release series. With the addition of python 3.8 (#3719), this version is now compatible with all current python versions that are not end-of-life:

  • 3.5

  • 3.6

  • 3.7

  • 3.8

Features#

  • Add the AiiDA Graph Explorer (AGE) a generic tool for traversing provenance graph [#3686]

  • Add the BaseRestartWorkChain which makes it easier to write a simple work chain wrapper around another process with automated error handling [#3748]

  • Add provenance_exclude_list attribute to CalcInfo data structure, allowing to prevent calculation input files from being permanently stored in the repository [#3720]

  • Add the verdi node repo dump command [#3623]

  • Add more methods to control cache invalidation of completed process node [#3637]

  • Allow documentation to be build without installing and configuring AiiDA [#3669]

  • Add option to expand namespaces in sphinx directive [#3631]

Performance#

  • Add node_type to list of immutable model fields, preventing repeated database hits [#3619]

  • Add cache for entry points in an entry point group [#3622]

  • Improve the performance when exporting many groups [#3681]

Changes#

  • CalcJob: move presubmit call from CalcJob.run to Waiting.execute [#3666]

  • CalcJob: do not pause when exception thrown in the presubmit [#3699]

  • Move CalcJob spec validator to corresponding namespaces [#3702]

  • Move getting completed job accounting to retrieve transport task [#3639]

  • Move last_job_info from JSON-serialized string to dictionary [#3651]

  • Improve SqlAlchemy session handling for QueryBuilder [#3708]

  • Use built-in open instead of io.open, which is possible now that python 2 is no longer supported [#3615]

  • Add non-zero exit code for verdi daemon status [#3729]

Bug fixes#

  • Deal with unreachable daemon worker in get_daemon_status [#3683]

  • Django backend: limit batch size for bulk_create operations [#3713]

  • Make sure that datetime conversions ignore None [#3628]

  • Allow empty key_filename in verdi computer configure ssh and reuse cooldown time when reconfiguring [#3636]

  • Update pyyaml to v5.1.2 to prevent arbitrary code execution [#3675]

  • QueryBuilder: fix validation bug and improve message for in operator [#3682]

  • Consider ‘AIIDA_TEST_PROFILE’ in ‘get_test_backend_name’ [#3685]

  • Ensure correct types for QueryBuilder().dict() with multiple projections [#3695]

  • Make local modules importable when running verdi run [#3700]

  • Fix bug in upload_calculation for CalcJobs with local codes [#3707]

  • Add imports from urllib to dbimporters [#3704]

Developers#

  • Moved continuous integration from Travis to Github actions [#3571]

  • Replace custom unit test framework for pytest and move all tests to tests top level directory [#3653][#3674][#3715]

  • Cleaned up direct dependencies and relaxed requirements where possible [#3597]

  • Set job poll interval to zero in localhost pytest fixture [#3605]

  • Make command line deprecation warnings visible with test profile [#3665]

  • Add docker image with minimal running AiiDA instance [#3722]

v1.0.1#

Improvements#

  • Improve the backup mechanism of the configuration file: unique backup written at each update [#3581]

  • Forward verdi code delete to verdi node delete [#3546]

  • Homogenize and improve output of verdi computer test [#3544]

  • Scheduler SLURM: support UNLIMITED and NOT_SET as values for requested walltimes [#3586]

  • Set default for the safe_interval option of verdi computer configure [#3590]

  • Create backup of configuration file before migrating [#3568]

  • Add python_requires to setup.json necessary for future dropping of python 2 [#3574]

  • Remove unused QB methods/functions [#3526]

  • Move pgtest argument of TemporaryProfileManager to constructor [#3486]

  • Add filename argument to SinglefileData constructor [#3517]

  • Mention machine in SSH connection exception message [#3536]

  • Docs: Expand on QB order_by information [#3548]

  • Replace deprecated pymatgen site.species_and_occu with site.species [#3480]

  • QueryBuilder: add deepcopy implementation and queryhelp property [#3524]

Bug fixes#

  • Fix verdi calcjob gotocomputer when key_filename is missing [#3593]

  • Fix bug in database migrations where schema generation determination excepts for old databases [#3582]

  • Fix false positive for verdi database integrity detect-invalid-links [#3591]

  • Config migration: handle edge case where daemon key is missing from daemon_profiles [#3585]

  • Raise when unable to detect name of local timezone [#3576]

  • Fix bug for CalcJob dry runs with store_provenance=False [#3513]

  • Migrations for legacy and now illegal default link label _return, export version upped to 0.8 [#3561]

  • Fix REST API attributes_filter and extras_filter [#3556]

  • Fix bug in plugin Factory classes for python 3.7 [#3552]

  • Make PolishWorkChains checkpointable [#3532]

  • REST API: fix generator of full node namespace [#3516]

v1.0.0#

Overview of changes#

The following is a summary of the major changes and improvements from v0.12.* to v1.0.0.

  • Faster workflow engine: the new message-based engine powered by RabbitMQ supports tens of thousands of processes per hour and greatly speeds up workflow testing. You can now run one daemon per AiiDA profile.

  • Faster database queries: the switch to JSONB for node attributes and extras greatly improves query speed and reduces storage size by orders of magnitude.

  • Robust calculations: AiiDA now deals with network connection issues (automatic retries with backoff mechanism, connection pooling, …) out of the box. Workflows and calculations are all Processes and can be “paused” and “played” anytime.

  • Better verdi commands: the move to the click framework brings homogenous command line options across all commands (loading nodes, …). You can easily add new commands through plugins.

  • Easier workflow development: Input and output namespaces, reusing specs of sub-processes and less boilerplate code simplify writing WorkChains and CalcJobs, while also enabling powerful auto-documentation features.

  • Mature provenance model: Clear separation between data provenance (Calculations, Data) and logical provenance (Workflows). Old databases can be migrated to the new model automatically.

  • python3 compatible: AiiDA 1.0 is compatible with both python 2.7 and python 3.6 (and later). Python 2 support will be dropped in the coming months.

Detailed list of changes#

Below a (non-exhaustive) list of changes by category. Changes between 1.0 alpha/beta releases are not included - for those see the changelog of the corresponding releases.

Engine and daemon#

  • Implement the concept of an “exit status” for all calculations, allowing a programmatic definition of success or failure for all processes [#1189]

  • All calculations now go through the Process layer, homogenizing the state of work and job calculations [#1125]

  • Allow None as default for arguments of process functions [#2582]

  • Implement the new calcfunction decorator. [#2203]

  • Each profile now has its own daemon that can be run completely independently in parallel [#1217]

  • Polling based daemon has been replaced with a much faster event-based daemon [#1067]

  • Replaced Celery with Circus as the daemonizer of the daemon [#1213]

  • The daemon can now be stopped without loading the database, making it possible to stop it even if the database version does not match the code [#1231]

  • Implement exponential backoff retry mechanism for transport tasks [#1837]

  • Pause CalcJob when transport task falls through exponential backoff [#1903]

  • Separate CalcJob submit task in folder upload and scheduler submit [#1946]

  • Each daemon worker now respects an optional minimum scheduler polling interval [#1929]

  • Make the execmanager.retrieve_calculation idempotent’ish [#3142]

  • Make the execmanager.upload_calculation idempotent’ish [#3146]

  • Make the execmanager.submit_calculation idempotent’ish [#3188]

  • Implement a PluginVersionProvider for processes to automatically add versions of aiida-core and plugin to process nodes [#3131]

Processes#

  • Implement the ProcessBuilder which simplifies the definition of Process inputs and the launching of a Process [#1116]

  • Namespaces added to the port containers of the ProcessSpec class [#1099]

  • Convention of leading underscores for non-storable inputs is replaced with a proper non_db attribute of the Port class [#1105]

  • Implement a Sphinx extension for the WorkChain class to automatically generate documentation from the workchain definition [#1155]

  • WorkChains can now expose the inputs and outputs of another WorkChain, which is great for writing modular workflows [#1170]

  • Add built-in support and API for exit codes in WorkChains [#1640], [#1704], [#1681]

  • Implement method for CalcJobNode to create a restart builder [#1962]

  • Add CalculationTools base and entry point aiida.tools.calculations [#2331]

  • Generalize Sphinx workchain extension to processes [#3314]

  • Collapsible namespace in sphinxext [#3441]

  • The retrieve_singlefile_list has been deprecated and is replaced by retrieve_temporary_list [#3041]

  • Automatically set CalcInfo.uuid in CalcJob.run [#2874]

  • Allow the usage of lambda functions for InputPort default values [#3465]

ORM#

  • Implementat AuthInfo class which allows custom configuration per configured computer [#1184]

  • Add efficient count method for aiida.orm.groups.Group [#2567]

  • Speed up creation of Nodes in the AiiDA ORM [#2214]

  • Enable use of tuple in QueryBuilder.append for all ORM classes [#1608], [#1607]

  • Refactor the ORM to have explicit front-end and back-end layers [#2190][#2210][#2225][#2227][#2481]

  • Add support for indexing and slicing in orm.Group.nodes iterator [#2371]

  • Add support for process classes to QueryBuilder.append [#2421]

  • Change type of uuids returned by the QueryBuilder to unicode [#2259]

  • The AttributeDict is now constructed recursively for nested dictionaries [#3005]

  • Ensure immutability of CalcJobNode hash before and after storing [#3130]

  • Fix bug in the RemoteData._clean method [#1847]

  • Fix bug in QueryBuilder.first() for multiple projections [#2824]

  • Fix bug in delete_nodes when passing pks of non-existing nodes [#2440]

  • Remove unserializable data from metadata in Log records [#2469]

Data#

  • Fix bug in parse_formula for formulas with leading or trailing whitespace [#2186]

  • Refactor Orbital code and fix some bugs [#2737]

  • Fix bug in the store method of CifData which would raise an exception when called more than once [#1136]

  • Allow passing directory path in FolderData constructor [#3359]

  • Add element X to the elements list in order to support unknown species [#1613]

  • Various bug and consistency fixes for CifData and StructureData [#2374]

  • Changes to Data class attributes and TrajectoryData data storage [#2310][#2422]

  • Rename ParameterData to Dict [#2530]

  • Remove the FrozenDict data sub class [#2532]

  • Remove the Error data sub class [#2529]

  • Make Code a real sub class of Data [#2193]

  • Implement the has_atomic_sites and has_unknown_species properties for the CifData class [#1257]

  • Change default library used in _get_aiida_structure (converting CifData to StructureData) from ase to pymatgen [#1257]

  • Add converter for UpfData from UPF to JSON format [#3308]

  • Fix potential inefficiency in aiida.tools.data.cif converters [#3098]

  • Fix bug in KpointsData.reciprocal_cell() [#2779]

  • Improve robustness of parsing versions and element names from UPF files [#2296]

Verdi command line interface#

  • Migrate verdi to the click infrastructure [#1795]

  • Add a default user to AiiDA configuration, eliminating the need to retype user information for every new profile [#2734]

  • Implement tab-completion for profile in the -p option of verdi [#2345]

  • Homogenize the interface of verdi quicksetup and verdi setup [#1797]

  • Add the option --version to verdi to display current version [#1811]

  • verdi computer configure can now read inputs from a yaml file through the --config option [#2951]

External database importers#

  • Add importer class for the Materials Platform of Data Science API, which hosts the Pauling file data [#1238]

  • Add an importer class for the Materials Project API [#2097]

Database#

  • Add an index to columns of DbLink for SqlAlchemy [#2561]

  • Creating unique constraint and indexes at the db_dbgroup_dbnodes table for SqlAlchemy [#1680]

  • Performance improvement for adding nodes to group [#1677]

  • Make UUID columns unique in SqlAlchemy [#2323]

  • Allow PostgreSQL connections via unix sockets [#1721]

  • Drop the unused nodeversion and public columns from the node table [#2937]

  • Drop various unused columns from the user table [#2944]

  • Drop the unused transport_params column from the computer table [#2946]

  • Drop the DbCalcState table [#2198]

  • [Django]: migrate the node attribute and extra schema to use JSONB, greatly improving storage and querying efficiency [#3090]

  • [SqlAlchemy]: Improve speed of node attribute and extra deserialization [#3090]

Export and Import#

  • Implement the exporting and importing of node extras [#2416]

  • Implement the exporting and importing of comments [#2413]

  • Implement the exporting and importing of logs [#2393]

  • Add export_parameters to the metadata.json in archive files [#3386]

  • Simplify the data format of export archives, greatly reducing file size [#3090]

  • verdi import automatically migrates archive files of old formats [#2820]

Miscellaneous#

  • Refactor unit test managers and add basic fixtures for pytest [#3319]

  • REST API v4: updates to conform with aiida-core==1.0.0 [#3429]

  • Improve decorators using the wrapt library such that function signatures are properly maintained [#2991]

  • Allow empty enabled and disabled keys in caching configuration [#3330]

  • AiiDA now enforces UTF-8 encoding for text output in its files and databases. [#2107]

Backwards-incompatible changes (only a sub-set)#

  • Remove aiida.tests and obsolete aiida.storage.tests.test_parsers entry point group [#2778]

  • Implement new link types [#2220]

  • Rename the type strings of Groups and change the attributes name and type to label and type_string [#2329]

  • Make various protected Node methods public [#2544]

  • Rename DbNode.type to DbNode.node_type [#2552]

  • Rename the ORM classes for Node sub classes JobCalculation, WorkCalculation, InlineCalculation and FunctionCalculation [#2184][#2189][#2192][#2195][#2201]

  • Do not allow the copy or deepcopy of Node, except for Data nodes [#1705]

  • Remove aiida.control and aiida.utils top-level modules; reorganize aiida.common, aiida.manage and aiida.tools [#2357]

  • Make the node repository API backend agnostic [#2506]

  • Redesign the Parser class [#2397]

  • [Django]: Remove support for datetime objects from node attributes and extras [#3090]

  • Enforce specific precision in clean_value for floats when computing a node’s hash [#3108]

  • Move physical constants from aiida.common.constants to external qe-tools package [#3278]

  • Add type checks to all plugin factories [#3456]

  • Disallow pickle when storing numpy array in ArrayData [#3434]

  • Remove implementation of legacy workflows [#2379]

  • Implement CalcJob process class that replaces the deprecated JobCalculation [#2389]

  • Change the structure of the CalcInfo.local_copy_list [#2581]

  • QueryBuilder: Change ‘ancestor_of’/’descendant_of’ to ‘with_descendants’/’with_ancestors’ [#2278]

v0.12.4#

Improvements#

  • Added new endpoint in rest api to get list of distinct node types [#2745]

  • Travis: port the deploy stage from the development branch [#2816]

Minor bug fixes#

  • Corrected the graph export set expansion rules [#2632]

Miscellaneous#

  • Backport streamlined quick install instructions from provenance_redesign [#2555]

  • Remove useless chainmap dependency [#2799]

  • Add aiida-core version to docs home page [#3058]

  • Docs: add note on increasing work_mem [#2952]

v0.12.3#

Improvements#

  • Fast addition of nodes to groups with skip_orm=True [#2471]

  • Add environment.yml for installing dependencies using conda; release of aiida-core on conda-forge channel [#2081]

  • REST API: io tree response now includes link type and node label [#2033] [#2511]

  • Backport postgres improvements for quicksetup [#2433]

  • Backport aiida.get_strict_version (for plugin development) [#2099]

Minor bug fixes#

  • Fix security vulnerability by upgrading paramiko to 2.4.2 [#2043]

  • Disable caching for inline calculations (broken since move to workfunction-based implementation) [#1872]

  • Let verdi help return exit status 0 [#2434]

  • Decode dict keys only if strings (backport) [#2436]

  • Remove broken verdi-plug entry point [#2356]

  • verdi node delete (without arguments) no longer tries to delete all nodes [#2545]

  • Fix plotting of BandsData objects [#2492]

Miscellaneous#

  • REST API: add tests for random sorting list entries of same type [#2106]

  • Add various badges to README [#1969]

  • Minor documentation improvements [#1955]

  • Add license file to MANIFEST [#2339]

  • Add instructions when verdi import fails [#2420]

v0.12.2#

Improvements#

  • Support the hashing of uuid.UUID types by registering a hashing function [#1861]

  • Add documentation on plugin cutter [#1904]

Minor bug fixes#

  • Make exported graphs consistent with the current node and link hierarchy definition [#1764]

  • Fix link import problem under SQLA [#1769]

  • Fix cache folder copying [#1746] [1752]

  • Fix bug in mixins.py when copying node [#1743]

  • Fix pgtest failures (release-branch) on travis [#1736]

  • Fix plugin: return testrunner result to fail on travis, when tests don’t pass [#1676]

Miscellaneous#

  • Remove pycrypto dependency, as it was found to have sercurity flaws [#1754]

  • Set xsf as default format for structures visualization [#1756]

  • Delete unused utils/create_requirements.py file [#1702]

v0.12.1#

Improvements#

  • Always use a bash login shell to execute all remote SSH commands, overriding any system default shell [#1502]

  • Reduced the size of the distributed package by almost half by removing test fixtures and generating the data on the fly [#1645]

  • Removed the explicit dependency upper limit for scipy [#1492]

  • Resolved various dependency requirement conflicts [#1488]

Minor bug fixes#

  • Fixed a bug in verdi node delete that would throw an exception for certain cases [#1564]

  • Fixed a bug in the cif endpoint of the REST API [#1490]

v0.12.0#

Improvements#

  • Hashing, caching and fast-forwarding [#652]

  • Calculation no longer stores full source file [#1082]

  • Delete nodes via verdi node delete [#1083]

  • Import structures using ASE [#1085]

  • StructureData - pymatgen - StructureData roundtrip works for arbitrary kind names [#1285] [#1306] [#1357]

  • Output format of archive file can now be defined for verdi export migrate [#1383]

  • Automatic reporting of code coverage by unit tests has been added [#1422]

Critical bug fixes#

  • Add parser_name JobProcess options [#1118]

  • Node attribute reads were not always up to date across interpreters for SqlAlchemy [#1379]

Minor bug fixes#

  • Cell vectors not printed correctly [#1087]

  • Fix read-the-docs issues [#1120] [#1143]

  • Fix structure/band visualization in REST API [#1167] [#1182]

  • Fix verdi work list test [#1286]

  • Fix _inline_to_standalone_script in TCODExporter [#1351]

  • Updated reentry to fix various small bugs related to plugin registering [#1440]

Miscellaneous#

  • Bump qe-tools version [#1090]

  • Document link types [#1174]

  • Switch to trusty + postgres 9.5 on Travis [#1180]

  • Use raw SQL in sqlalchemy migration of Code [#1291]

  • Document querying of list attributes [#1326]

  • Document running aiida as a daemon service [#1445]

  • Document that Torque and LoadLever schedulers are now fully supported [#1447]

  • Cookbook: how to check the number of queued/running jobs in the scheduler [#1349]

v0.11.4#

Improvements#

  • PyCifRW upgraded to 4.2.1 [#1073]

Critical bug fixes#

  • Persist and load parsed workchain inputs and do not recreate to avoid creating duplicates for default inputs [#1362]

  • Serialize WorkChain context before persisting [#1354]

v0.11.3#

Improvements#

  • Documentation: AiiDA now has an automatically generated and complete API documentation (using sphinx-apidoc) [#1330]

  • Add JSON schema for connection of REST API to Materials Cloud Explore interface [#1336]

Critical bug fixes#

  • FINISHED_KEY and FAILED_KEY variables were not known to AbstractCalculation [#1314]

Minor bug fixes#

  • Make ‘REST’ extra lowercase, such that one can do pip install aiida-core[rest] [#1328]

  • CifData /visualization endpoint was not returning data [#1328]

Deprecations#

  • QueryTool (was deprecated in favor of QueryBuilder since v0.8.0) [#1330]

Miscellaneous#

  • Add gource config for generating a video of development history [#1337]

v0.11.2:#

Critical bug fixes#

  • Link types were not respected in Node.get_inputs for SqlAlchemy [#1271]

v0.11.1:#

Improvements#

  • Support visualization of structures and cif files with VESTA [#1093]

  • Better fallback when node class is not available [#1185]

  • CifData now supports faster parsing and lazy loading [#1190]

  • REST endpoint for CifData, API reports full list of available endpoints [#1228]

  • Various smaller improvements [#1100] [#1182]

Critical bug fixes#

  • Restore attribute immutability in nodes [#1111]

  • Fix daemonization issue that could cause aiida daemon to be killed [#1246]

Minor bug fixes#

v0.11.0:#

Improvements#

Core entities#

  • Computer: the shebang line is now customizable [#940]

  • KpointsData: deprecate buggy legacy implementation of k-point generation in favor of Seekpath [#1015]

  • Dict: to_aiida_type used on dictionaries now automatically converted to Dict [#947]

  • JobCalculation: parsers can now specify files that are retrieved locally for parsing, but only temporarily, as they are deleted after parsing is completed [#886] [#894]

Plugins#

  • Plugin data hooks: plugins can now add custom commands to verdi data [#993]

  • Plugin fixtures: simple-to-use decorators for writing tests of plugins [#716] [#865]

  • Plugin development: no longer swallow ImportError exception during import of plugins [#1029]

Verdi#

  • verdi shell: improve tab completion of imports in [#1008]

  • verdi work list: projections for verdi work list [#847]

Miscellaneous#

  • Supervisor removal: dependency on unix-only supervisor package removed [#790]

  • REST API: add server info endpoint, structure endpoint can return different file formats [#878]

  • REST API: update endpoints for structure visualization, calculation (includes retrieved input & output list), add endpoints for UpfData and more [#977] [#991]

  • Tests using daemon run faster [#870]

  • Documentation: updated outdated workflow examples [#948]

  • Documentation: updated import/export [#994],

  • Documentation: plugin quickstart [#996],

  • Documentation: parser example [#1003]

Minor bug fixes#

  • Fix bug with repository on external hard drive [#982]

  • Fix bug in configuration of pre-commit hooks [#863]

  • Fix and improve plugin loader tests [#1025]

  • Fix broken celery logging [#1033]

Deprecations#

  • async from aiida.work.run has been deprecated because it can lead to race conditions and thereby unexpected behavior [#1040]

v0.10.1:#

Improvements#

  • Improved exception handling for loading db tests [#968]

  • verdi work kill on workchains: skip calculation if it cannot be killed, rather than stopping [#980]

  • Remove unnecessary INFO messages of Alembic for SQLAlchemy backend [#1012]

  • Add filter to suppress unnecessary log messages during testing [#1014]

Critical bug fixes#

  • Fix bug in verdi quicksetup on Ubuntu 16.04 and add regression tests to catch similar problems in the future [#976]

  • Fix bug in verdi data list commands for SQLAlchemy backend [#1007]

Minor bug fixes#

  • Various bug fixes related to workflows for the SQLAlchemy backend [#952] [#960]

v0.10.0:#

Major changes#

  • The DbPath table has been removed and replaced with a dynamic transitive closure, because, among others, nested workchains could lead to the DbPath table exploding in size

  • Code plugins have been removed from aiida-core and have been migrated to their own respective plugin repositories and can be found here:

    Each can be installed from pip using e.g. pip install aiida-quantumespresso. Existing installations will require a migration (see update instructions in the documentation). For a complete overview of available plugins you can visit the registry.

Improvements#

  • A new entry retrieve_temporary_list in CalcInfo allows to retrieve files temporarily for parsing, while not having to store them permanently in the repository [#903]

  • New verdi command: verdi work kill to kill running workchains [#821]

  • New verdi command: verdi data remote [ls,cat,show] to inspect the contents of RemoteData objects [#743]

  • New verdi command: verdi export migrate allows the migration of existing export archives to new formats [#781]

  • New verdi command: verdi profile delete [#606]

  • Implemented a new option -m for the verdi work report command to limit the number of nested levels to be printed [#888]

  • Added a running field to the output of verdi work list to give the current state of the workchains [#888]

  • Implemented faster query to obtain database statistics [#738]

  • Added testing for automatic SqlAlchemy database migrations through alembic [#834]

  • Exceptions that are triggered in steps of a WorkChain are now properly logged to the Node making them visible through verdi work report [#908]

Critical bug fixes#

  • Export will now write the link types to the archive and import will properly recreate the link [#760]

  • Fix bug in workchain persistence that would lead to crashed workchains under certain conditions being resubmitted [#728]

  • Fix bug in the pickling of WorkChain instances containing an _if logical block in the outline [#904]

Minor bug fixes#

  • The logger for subclasses of AbstractNode is now properly namespaced to aiida. such that it works in plugins outside of the aiida-core source tree [#897]

  • Fixed a problem with the states of the direct scheduler that was causing the daemon process to hang during submission [#879]

  • Various bug fixes related to the old workflows in combination with the SqlAlchemy backend [#889] [#898]

  • Fixed bug in TCODexporter [#761]

  • verdi profile delete now respects the configured dbport setting [#713]

  • Restore correct help text for verdi --help [#704]

  • Fixed query in the ICSD importer element that caused certain structures to be erroneously skipped [#690]

Miscellaneous#

v0.9.1:#

Critical bug fixes#

  • Workchain steps will no longer be executed multiple times due to process pickles not being locked

Minor bug fixes#

  • Fix arithmetic operations for basic numeric types

  • Fixed verdi calculation cleanworkdir after changes in QueryBuilder syntax

  • Fixed verdi calculation logshow exception when called for WorkCalculation nodes

  • Fixed verdi import for SQLAlchemy profiles

  • Fixed bug in reentry and update dependency requirement to v1.0.2

  • Made octal literal string compatible with python 3

  • Fixed broken import in the ASE plugin

Improvements#

  • verdi calculation show now properly distinguishes between WorkCalculation and JobCalculation nodes

  • Improved error handling in verdi setup --non-interactive

  • Disable unnecessary console logging for tests

v0.9.0#

Data export functionality#

  • A number of new functionalities have been added to export band structures to a number of formats, including: gnuplot, matplotlib (both to export a python file, and directly PNG or PDF; both with support of LaTeX typesetting and not); JSON; improved agr (xmgrace) output. Also support for two-color bands for collinear magnetic systems. Added also possibility to specify export-format-specific parameters.

  • Added method get_export_formats() to know available export formats for a given data subclass

  • Added label prettifiers to properly typeset high-symmetry k-point labels for different formats (simple/old format, seekpath, …) into a number of plotting codes (xmgrace, gnuplot, latex, …)

  • Improvement of command-line export functionality (more options, possibility to write directly to file, possibility to pass custom options to exporter, by removing its DbPath dependency)

Workchains#

  • Crucial bug fix: workchains can now be run through the daemon, i.e. by using aiida.work.submit

  • Enhancement: added an abort and abort_nowait method to WorkChain which allows to abort the workchain at the earliest possible moment

  • Enhancement: added the report method to WorkChain, which allows a workchain developer to log messages to the database

  • Enhancement: added command verdi work report which for a given pk returns the messages logged for a WorkChain through the report method

  • Enhancement: workchain inputs ports with a valid default specified no longer require to explicitly set required=False but is overriden automatically

New plugin system#

  • New plugin system implemented, allowing to load aiida entrypoints, and working in parallel with old system (still experimental, though - command line entry points are not fully implemented yet)

  • Support for the plugin registry

Code refactoring#

  • Refactoring of Node to move as much as possible of the caching code into the abstract class

  • Refactoring of Data nodes to have the export code in the topmost class, and to make it more general also for formats exporting more than one file

  • Refactoring of a number of Data subclasses to support the new export API

  • Refactoring of BandsData to have export code not specific to xmgrace or a given format, and to make it more general

Documentation#

  • General improvements to documentation

  • Added documentation to upgrade AiiDA from v0.8.0 to v0.9.0

  • Added documentation of new plugin system and tutorial

  • Added more in-depth documentation on how to export data nodes to various formats

  • Added explanation on how to export band structures and available formats

  • Added documentation on how to run tests in developer’s guide

  • Documented Latex requirements

  • Updated WorkChain documentation for WaitingEquationOfState example

  • Updated AiiDA installation documentation for installing virtual environment

  • Updated documentation to use Jupyter

Enhancements#

  • Speedups the travis builds process by caching pip files between runs

  • Node can be loaded by passing the start of its UUID

  • Handled invalid verdi command line arguments; added help texts for same

  • upgraded Paramiko to 2.1.2 and avoided to create empty file when remote connection is failed

  • verdi calculation kill command is now available for SGE plugin

  • Updated Plum from 0.7.8 to 0.7.9 to create a workchain inputs that had default value and evaluated to false

  • Now QueryBuilder will be imported by default for all verdi commands

Bug Fixes#

  • Bug fixes in QE input parser

  • Code.get() method accepts the pk in integer or string format whereas Code.get_from_string() method accepts pk only in string format

  • verdi code show command now shows the description of the code

  • Bug fix to check if computer is properly configured before submitting the calculation

Miscellaneous#

  • Replacing dependency from old unmantained pyspglib to new spglib

  • Accept BaseTypes as attributes/extras, and convert them automatically to their value. In this way, for instance, it is now possible to pass a Int, Float, Str, … as value of a dictionary, and store all into a Dict.

  • Switch from pkg_resources to reentry to allow for much faster loading of modules when possible, and therefore allowing for good speed for bash completion

  • Removed obsolete code for Sqlite

  • Removed mayavi2 package from dependencies

v0.8.1#

Exporters#

  • Upgraded the TCODExporter to produce CIF files, conforming to the newest (as of 2017-04-26) version of cif_tcod.dic.

General#

  • Added dependency on six to properly re-raise exceptions

v0.8.0#

Installation and setup#

  • Simplified installation procedure by adopting standard python package installation method through setuptools and pip

  • Verdi install replaced by verdi setup

  • New verdi command quicksetup to simplify the setup procedure

  • Significantly updated and improved the installation documentation

General#

  • Significantly increased test coverage and implemented for both backends

  • Activated continuous integration through Travis CI

  • Application-wide logging is now abstracted and implemented for all backends

  • Added a REST API layer with hook through verdi cli: verdi restapi

  • Improved QueryBuilder

    • Composition model instead of inheritance removing the requirement of determining the implementation on import

    • Added keyword with_dbpath that makes QueryBuilder switch between using the DbPathand not using it.

    • Updated and improved documentation

  • The QueryTool as well as the class Node.query() method are now deprecated in favor of the QueryBuilder

  • Migration of verdi cli to use the QueryBuilder in order to support both database backends

  • Added option --project to verdi calculation list to specify which attributes to print

Documentation#

  • Documentation is restructured to improve navigability

  • Added pseudopotential tutorial

Database#

  • Dropped support for MySQL and SQLite to fully support efficient features in Postgres like JSONB fields

  • Database efficiency improvements with orders of magnitude speedup for large databases [added indices for daemon queries and node UUID queries]

  • Replace deprecated commit_on_success with atomic for Django transactions

  • Change of how SQLAlchemy internally uses the session and the engine to work also with forks (e.g. in celery)

Workflows#

  • Finalized the naming for the new workflow system from workflows2 to work

    • FragmentedWorkFunction is replaced by WorkChain

    • InlineCalculation is replaced by Workfunction

    • ProcessCalculation is replaced by WorkCalculation

  • Old style Workflows can still be called and run from a new style WorkChain

  • Major improvements to the WorkChain and Workfunction implementation

  • Improvements to WorkChain

    • Implemented a return statement for WorkChain specification

    • Logging to the database implemented through WorkChain.report() for debugging

  • Improved kill command for old-style workflows to avoid steps to remaing in running state

Plugins#

  • Added finer granularity for parsing PW timers in output

  • New Quantum ESPRESSO and scheduler plugins contributed from EPFL

    • ASE/GPAW plugins: Andrea Cepellotti (EPFL and Berkeley)

    • Quantum ESPRESSO DOS, Projwfc: Daniel Marchand (EPFL and McGill)

    • Quantum ESPRESSO phonon, matdyn, q2r, force constants plugins: Giovanni Pizzi, Nicolas Mounet (EPFL); Andrea Cepellotti (EPFL and Berkeley)

    • Quantum ESPRESSO cp.x plugin: Giovanni Pizzi (EPFL)

    • Quantum ESPRESSO neb.x plugin: Marco Gibertini (EPFL)

    • LSF scheduler: Nicolas Mounet (EPFL)

  • Implemented functionality to export and visualize molecular dynamics trajectories (using e.g. matplotlib, mayavi)

  • Improved the TCODExporter (some fixes to adapt to changes of external libraries, added some additional TCOD CIF tags, various bugfixes)

Various fixes and improvements#

  • Fix for the direct scheduler on Mac OS X

  • Fix for the import of computers with name collisions

  • Generated backup scripts are now made profile specific and saved as start_backup_<profile>.py

  • Fix for the vary_rounds warning

v0.7.1#

Functionalities#

  • Implemented support for Kerberos authentication in the ssh transport plugin.

  • Added _get_submit_script_footer to scheduler base class.

  • Improvements of the SLURM scheduler plugin.

  • Fully functional parsers for Quantumespresso CP and PW.

  • Better parsing of atomic species from PW output.

  • Array classes for projection & xy, and changes in kpoints class.

  • Added codespecific tools for Quantumespresso.

  • verdi code listnow shows local codes too.

  • verdi export can now export non user-defined groups (from their pk).

Fixes#

  • Fixed bugs in (old) workflow manager and daemon.

  • Improvements of the efficiency of the (old) workflow manager.

  • Fixed JobCalculation text prepend with multiple codes.

v0.7.0#

This release introduces a lot and significant changes & enhancements.

We worked on our new backend and now AiiDA can be installed using SQLAlchemy too. Many of the verdi commands and functionalities have been tested and are working with this backend. The full JSON support provided by SQLAlchemy and the latest versions of PostgreSQL enable significant speed increase in attribute related queries. SQLAlchemy backend choice is a beta option since some last functionalities and commands need to be implemented or improved for this backend. Scripts are provided for the transition of databases from Django backend to SQLAlchemy backend.

In this release we have included a new querying tool called QueryBuilder. It is a powerfull tool allowing users to write complex graph queries to explore the AiiDA graph database. It provides various features like selection of entity properties, filtering of results, combination of entities on specific properties as well as various ways to obtain the final result. It also provides the users an abstract way to query their data without enforcing them to write backend dependent queries.

Last but not least we have included a new workflow engine (in beta version) which is available through the verdi workflow2 command. The new workflows are easier to write (it is as close as writing python as possible), there is seamless mixing of short running tasks with long running (remote) tasks and they encourage users to write reusable workflows. Moreover, debugging of workflows has been made easier and it is possible both in-IDE and through logging.

List of changes:#

  • Installation procedure works with SQLAlchemy backend too (SQLAlchemy option is still in beta).

  • Most of the verdi commands work with SQLAlchemy backend.

  • Transition script from Django schema of version 0.7.0 to SQLAlchemy schema of version 0.7.0.

  • AiiDA daemon redesigned and working with both backends (Django & SQLAlchemy).

  • Introducing new workflow engine that allows better debugging and easier to write workflows. It is available under the verdi workflows2 command. Examples are also added.

  • Old workflows are still supported and available under the “verdi workflow” command.

  • Introducing new querying tool (called QueryBuilder). It allows to easily write complex graph queries that will be executed on the AiiDA graph database. Extensive documentation also added.

  • Unifying behaviour of verdi commands in both backends.

  • Upped to version 0.4.2 of plum (needed for workflows2)

  • Implemented the validator and input helper for Quantum ESPRESSO pw.x.

  • Improved the documentation for the pw (and cp) input plugins (for all the flags in the Settings node).

  • Fixed a wrong behavior in the QE pw/cp plugins when checking for the parser options and checking if there were further unknown flags in the Settings node. However, this does not solve yet completely the problem (see issue #219).

  • Implemented validator and input helper for Quantum ESPRESSO pw.x.

  • Added elements with Z=104-112, 114 and 116, in aiida.common.constants.

  • Added method set_kpoints_mesh_from_density in KpointsData class.

  • Improved incremental backup documentation.

  • Added backup related tests.

  • Added an option to test_pw.py to run also in serial.

  • SSH transport, to connect to remote computers via SSH/SFTP.

  • Support for the SGE and SLURM schedulers.

  • Support for Quantum ESPRESSO Car-Parrinello calculations.

  • Support for data nodes to store electronic bands, phonon dispersion and generally arrays defined over the Brillouin zone.

v0.6.0#

We performed a lot of changes to introduce in one of our following releases a second object-relational mapper (we will refer to it as back-end) for the management of the used DBMSs and more specifically of PostgreSQL. SQLAlchemy and the latest version of PostgreSQL allows AiiDA to store JSON documents directly to the database and also to query them. Moreover the JSON query optimization is left to the database including also the use of the JSON specific indexes. There was major code restructuring to accommodate the new back-end resulting to abstracting many classes of the orm package of AiiDA.

Even if most of the needed restructuring & code addition has been finished, a bit of more work is needed. Therefore even in this version, Django is the only available back-end for the end user.

However, the users have to update their AiiDA configuration files by executing the migration file that can be found at YOUR_AIIDA_DIR/aiida/common/additions/migration.py as the Linux user that installed AiiDA in your system. (e.g. python YOUR_AIIDA_DIR/aiida/common/additions/migration.py)

List of changes:#

  • Back-end selection (Added backend selection). SQLAlchemy selection is disabled for the moment.

  • Migration scripts for the configuration files of AiiDA (SQLAlchemy support).

  • Enriched link description in the database (to enrich the provenance model).

  • Corrections for numpy array and cell. List will be used with cell.

  • Fixed backend import. Verdi commands load as late as possible the needed backend.

  • Abstraction of the basic AiiDA orm classes (like node, computer, data etc). This is needed to support different backends (e.g. Django and SQLAlchemy).

  • Fixes on the structure import from QE-input files.

  • SQLAlchemy and Django benchmarks.

  • UltraJSON support.

  • requirements.txt now also include SQLAlchemy and its dependencies.

  • Recursive way of loading JSON for SQLAlchemy.

  • Improved way of accessing calculations and workflows attached to a workflow step.

  • Added methods to programmatically create new codes and computers.

v0.5.0#

General#

  • Final paper published, ref: G. Pizzi, A. Cepellotti, R. Sabatini, N. Marzari, and B. Kozinsky, AiiDA: automated interactive infrastructure and database for computational science, Comp. Mat. Sci 111, 218-230 (2016)

  • Core, concrete, requirements kept in requirements.txt and optionals moved to optional_requirements.txt

  • Schema change to v1.0.2: got rid of calc_states.UNDETERMINED

Import/export, backup and code interaction#

  • [non-back-compatible] Now supporting multiple codes execution in the same submission script. Plugin interface changed, requires adaptation of the code plugins.

  • Added import support for XYZ files

  • Added support for van der Waals table in QE input

  • Restart QE calculations avoiding using scratch using copy of parent calc

  • Adding database importer for NNIN/C Pseudopotential Virtual Vault

  • Implemented conversion of pymatgen Molecule lists to AiiDA’s TrajectoryData

  • Adding a converter from pymatgen Molecule to AiiDA StructureData

  • Queries now much faster when exporting

  • Added an option to export a zip file

  • Added backup scripts for efficient incremental backup of large AiiDA repositories

API#

  • Added the possibility to add any kind of Django query in Group.query

  • Added TCOD (Theoretical Crystallography Open Database) importer and exporter

  • Added option to sort by a field in the query tool

  • Implemented selection of data nodes and calculations by group

  • Added NWChem plugin

  • Change default behaviour of symbolic link copy in the transport plugins: “put”/”get” methods -> symbolic links are followed before copy; “copy” methods -> symbolic links are not followed (copied “as is”).

Schedulers#

  • Explicit Torque support (some slightly different flags)

  • Improved PBSPro scheduler

  • Added new num_cores_per_machine and num_cores_per_mpiproc fields for pbs and torque schedulers (giving full support for MPI+OpenMP hybrid codes)

  • Direct scheduler added, allowing calculations to be run without batch system (i.e. directly call executable)

verdi#

  • Support for profiles added: it allows user to switch between database configurations using the verdi profile command

  • Added verdi data structure import --file file.xyz for importing XYZ

  • Added a verdi data upf exportfamily command (to export an upf pseudopotential family into a folder)

  • Added new functionalities to the verdi group command (show list of nodes, add and remove nodes from the command line)

  • Allowing verdi export command to take group PKs

  • Added ASE as a possible format for visualizing structures from command line

  • Added possibility to export trajectory data in xsf format

  • Added possibility to show trajectory data with xcrysden

  • Added filters on group name in verdi group list

  • Added possibility to load custom modules in the verdi shell (additional property verdishell.modules created; can be set with verdi devel setproperty verdishell.modules)

  • Added verdi data array show command, using json_date serialization to display the contents of ArrayData

  • Added verdi data trajectory deposit command line command

  • Added command options --computer and --code to verdi data * deposit

  • Added a command line option --all-users for verdi data * list to list objects, owned by all users

On this page
Edit on GitHub
Show Source