# AiiDA Team Meeting 2025-02-06 ###### tags: `team meetings` ###### time: 14:00 CET [TOC] ### Present - Alex - Julian - Edan - Giovanni - Xing - Jusong - Kristjan ### Catch-up round *Max. 3 minutes each* - Ali: (I'll not be there, so please read on my behalf) - Open PRs on `aiida-core` - [`Transport`: feat `compress` & `extract` methods](https://github.com/aiidateam/aiida-core/pull/6743) - [`RemoteStashCompressedData`: compression modes while stashing](https://github.com/aiidateam/aiida-core/pull/6746) - Benchmarking and communication with FirecREST team: - To download files below 5 MB in size - The new benchmark is really good! FirecREST v2 + async outperform both paramiko, and asyncssh. (although it's still slower than filesystem i/o implemtations like rsync. But still, faster for us who uses python libraries of SSH) - To download files above 5 MB in size - Could not test due to deployment bugs, FirecREST team is on the issue. - Suggested by Giovanni: Discuss if we have remove a group called 'Internal AiiDA team' from Google groups. Or in gerneral remove whatever we have from Google groups, including AiiDAlab, etc. --> Remove it since nobody uses it and we have enough means of communication. @Kristjan will do it - Alex: - Release v2.7.0 features (see [GitHub project](https://github.com/orgs/aiidateam/projects/15)): - async transport plugin - fully supported sqlite backend for querybuilder - profile mirroring - py3.13 support - Use pydantic to specify a schema for each ORM - process force kill - We create a release candidate on the 20th February with all features that we want to have for release. This means one can install it with `pip install --pre aiida-core`. Then we will leave it open for production tests for a month until we do an official release. Going from the release candidate to release we only add bugfixes, no new features will be added. My questions: How do we do production tests? Are there some specific aiida-quantumespresso workflows we can use for it? We need: - robustness tests for async transport - performance tests for pydantic serialization - sqlite vs postgress performance tests (add these performance tests to docs) -> For the production tests we could make a google docs - @edan aiidalab-qe app Silicon and gold for bands/PDOS - @xing aiidalab-qe app XPS/XAS - Alex asks @Michael@Timo@Marnik for the MC3D workflows (+test aiida-pseudo) - Alex asks @Natalia Wannier90 workflow (automatic PDWF Wannierization) - Alex asks @Miki phonon vibroscopy workfows and muon workflows (test also via AiiDAlab-QE) - @xing aiida-workgraph test suite (and aiida-shell) - @Julian happy to take over aiida-project - Alex asks @Michael@Marnik aiida-submission-controller - @Kristjan@Julian compare for reps api benchmarks of Ali [khsrali/aiida-profile (private repo)](https://github.com/khsrali/aiida-profiling) - Py3.13 support status - `circus` should work with python 3.13 (added CI [circus/1221](https://github.com/circus-tent/circus/pull/1221), PR open for new release [circus/1223](https://github.com/circus-tent/circus/pull/1223) - The `Container` from `disk-objectstore` is not correctly closing its connection resulting in open file descriptors that accumulate in the tests (see issue [aiida-core/6739](https://github.com/aiidateam/aiida-core/issues/6739)). The PRs [aiida-core/6741](https://github.com/aiidateam/aiida-core/pull/6741) and [disk-objectstore/179](https://github.com/aiidateam/disk-objectstore/pull/179) solve this issue but need some time as disk-objectstore needs some maintenance updates (CI runs on py3.8-py3.11) - Julian - CI currently fails due to mamba issue: [Failing CI due to use of deprecated `--no-python-version-warning` in `mamba`](https://github.com/aiidateam/aiida-core/issues/6745) - Class-scoped fixtures? Refactor fixture content to utility functions idea brought up by @Alex? Issue: [Refactor fixtures to allow for different scopes](https://github.com/aiidateam/aiida-core/issues/6738). Worth the effort? - Other advantage of utility functions: Able to use them outside of the fixtures - For testing profile mirroring, I created some fixtures that create nodes and groups atomically, but might still take too long (would like to share screen later if there is time). - Fixtures to create ArithmeticAdd CalcJobNode and MultiplyAdd WorkChainNodes always actually submit/run the Processes. Mock-fixture that manually constructs the node might make tests quicker. Draft PR for the ArithmeticAdd here: [Add `ArithmeticAdd` CJ Node fixture without `run` or `submit`](https://github.com/aiidateam/aiida-core/pull/6733/). - Test speed to see if it's actually worth it - Next week Tuesday 10 AM, discussion on Rabbull's changes to bulk update of node extras: [Storage: Add `extend_json` Tag for Enhanced JSON Field Handling in `bulk_update` Operations](https://github.com/aiidateam/aiida-core/pull/6659) (and other SQL-related changes he implemented). All welcome to join! - Edan - Would like to resolve the [pydantic PR](https://github.com/aiidateam/aiida-core/pull/6255) asap - Will test resolution to CLI issue once more for completeness - Benchmarking perhaps can wait (see Alex comments regarding pre-release) - Working on pydantic schemas for common workflows - Xing - WorkGraph coding week (02.24-02.28), with Alex and Julian. * Jason * NumFOCUS news for Feb. * Help plan and guide a NumFOCUS "Chemistry and Materials Science" event and/or group. You can find the [registration link](https://numfocus-org.zoom.us/meeting/register/FaLHt2oGQfWk2W0Xpu6bRQ). I'll forward the email to you if you are interested. * GSoC projects idea list (DDL is in 4 days) [Idea list](https://github.com/aiidateam/aiida-core/wiki/GSoC-2025-Projects) * The record of Julia workshop where I did AiiDA talk is [now online](https://www.youtube.com/playlist?list=PLagOIMNRoCsV1-BQAygx86cQt1l5Psh9Z) * Prototyping on how get rid of nest-asyncio with using greenlet can bring also the non-blocking calcfunction possible. https://github.com/unkcpz/plumpy/pull/39 * Showing how to use safe de/se loader in plumpy for message passing. https://github.com/aiidateam/plumpy/pull/310 Kristjan: - nothing to report