# AiiDA Team Meeting 2024-07-25

###### tags: `team meetings`
###### time: 14:00 CET

[TOC]

### Present

* Marnik
* Ali
* Alex
* Giovanni
* Seb

### Catch-up round

*Max. 3 minutes each*

Marnik:
* Hero run on Alps-Todi: quite successful, working with MARVEL communication team (Nicola Nosengo) on article to publish and disseminate. Engine stability was noticably improved VS previous hero runs.
* [EuroSciPy](https://euroscipy.org/2024/index.html) (26-30 Aug): [tutorial](https://pretalx.com/euroscipy-2024/talk/BYESWT/) and [talk](https://pretalx.com/euroscipy-2024/talk/HKVEXW/). Touch base next week on content?
* [Responsibilities](https://hackmd.io/hYaz4oXyTfWDUWKIKc0HUg): something we forgot to finalise during the coding week :sweat_smile:.

Ali:
Please have a look at the [semi digested results](https://hackmd.io/pn1ErwXIQperJhIcwVgm-w) of questionnaire, taken from the [full responses](https://docs.google.com/document/d/1KTag-gpNA0MbU_f3t7_ZDxK0Yf_1QNSlpAix17GLsJg/edit#heading=h.j8zz1788bpf).
Apart from all positive feedback, there are a few things needs our attention:
- Documentation needs extensive improvement, many comments on lack of discoverability and clearance. **+19 mentions**
- Setting up / debugging workflows/workchain. **+17 mentions** --> [WorkGraph in development]
- The querybuilder is difficult to use. **+7 mentions** [GP: Edan is working on a GUI]
- Some requested folder-based behaviour. **+6 mentions** --> [`verdi mirror`, git like behaviour]
- Possible issue on daemon performance. **+3 mentions** [is there some specific request? Note also recent issue by Julian on not waiting 2 mins for a single submission]
- Some miscellaneous suggestions (please check the document).

Giovanni:
* Reply to MARVEL SAB/IAB: https://drive.google.com/drive/folders/1wlNgMCIgewlW6tcON7qHt7b3vQoi-fPQ (I plan to send it end of Aug to the board)
* Issue opened with Julian on improving performance from 2 mins down to ~5 secs (maybe) for SSH transport, to avoid various waiting times if nothing happened recently
* two different appraoches: first 30 secs: keep track when the connection was last closed
* other 3x30s: keep the connection open for a few (~5? configurable?) more seconds, so it can be reused by tasks following just after
* Contributed to the fix of the locked DB in SQLite, see discussion by Seb. I hope is the main issue that actually happens, and with this, one can run in production with SQLite (together with the query extensions below)
* Note: would still be good to factor out the broker, is someone already working on this? (Seb, Davide G., ...?)
* I investigated how to get feature parity with SQLite (wrt SQLalchemy).
* I think I fixed the `has_key` issue: https://github.com/aiidateam/aiida-core/issues/6552 (no time to actually implement and test until the second half of August though... Feel free to pick it up)
* There is a probable solution for `contains`: https://github.com/aiidateam/aiida-core/issues/6494 but I need still to see how to implement this in SQLA inside QueryBuilder.
* With these two, I think all queries are covered?
* If so, I would benchmark speed (in general, for queries), and also (Ali, Kristjan, Alex) further investigate using the SQLite_zip backend for serving Materials Cloud APIs (benchmarking also this additional layer of going via the REST API via a web server)
* Discuss timing/plan for WorkGraph (and various tutorials until end of Aug)

Sebastiaan:
* **Merged PRs:**
* [SqliteDosStorage: Catch locked database when setting global variable ](https://github.com/aiidateam/aiida-core/pull/6534)
* **Open PRs:**
* [Engine: Ensure node is sealed when process excepts](https://github.com/aiidateam/aiida-core/pull/6549)
* Think it is time to soon make a patch release At least #6549 and #6551 need to go in. Anything else?

Alexander:
* [aiidateam/aiida-plugin-cutter/pull/108](https://github.com/aiidateam/aiida-plugin-cutter/pull/108) Update and modernize cookiecutter template using `hatch` recommended by PyPA. With this we get a unified interface to run all developer tools within isolated environments.
* [C2SM/WCFlow/pull/13](https://github.com/C2SM/WCFlow/pull/13) and [C2SM/WCFlow/pull/14](https://github.com/C2SM/WCFlow/pull/14) (sadly repo is private ATM) YAML parser for cyclic jobs using aiida-shell: Initialized proper repo structure and refactored our (Julian and me) vertical slice code to proper code

### Specific Agenda Items

* Responsabilities discussion: taken from document, work from Alex, adapted by MBx.

#### Responsibility list (Component -> Person)

* Storage-database+QB: Edan, Julian, Giovanni [SQL]
* Storage-repository: **Ali**, **Jason** [disk-objectstore], Giovanni [disk-objecstore]
* Website: Edan, Julian, Ali
* Documentation: Alexander, Marnik
* REST-API: Edan, Kristjan, Jason, Alexander
* verdi CLI: Alexander, Marnik, Miki
* DevOps: Jason, Marnik, Julian
* Data types: Miki, Julian, Edan
* Engine: Marnik, Jason, Julian, Ali
* Tests: Julian, Alexander
* AiiDA-Workgraph: Xing, Alexander, Julian
* Scheduler/Transport: Ali, Jason, Alexander, Giovanni
* Dependencies: Sebastiaan, Alexander
* Provenance Graph: Kristjan, Edan, Xing