# AiiDA Team Meeting 2022-12-14

###### tags: `team meetings`
###### time: 10am CET

[TOC]

### Present

* Marnik
* Giovanni
* santa
* Kristjan
* Francisco
* Jason
* Xing
* Sebastiaan

### Catch-up round

*Max. 3 minutes each*

Chris:
* typing stuff
* Stuff on workchains

Marnik:
* Ran "hero run" on LUMI-G. AiiDA performed admirably, but Slurm scheduler could not handle the rate of submissions.
* Moved default branch of `aiida-tutorials` to `main`. Also [added Quantum Mobile instructions](https://github.com/aiidateam/aiida-tutorials/pull/463) with link from [this PR](https://github.com/marvel-nccr/quantum-mobile/pull/200). Still working on adding previous versions at bottom of landing page.
* Highlighted Usability issue: verdi setup could automatically set up localhost [[#749](https://github.com/aiidateam/aiida-core/issues/749)].

Jason:
* Summary of the discussion AiiDAlab [package management system](https://github.com/aiidalab/aiidalab-docker-stack/issues/343#issuecomment-1341061375), "pip or conda".
* The way to set daint account without write it to prepend_text?
* CSCS MFA test: beta users to enroll 1st week of Jan. Till March lask week.
* We got more experience on the docker image for AiiDAlab, we need update the AiiDA image. The open question is [which base image we use?](https://github.com/aiidateam/aiida-core/discussions/5714#discussioncomment-4075303)

Francisco:
* Attending a ["Laboratory Automation Workshop"](https://www.epfl.ch/labs/create/lab-automation-workshop-2022/) this friday: managed to get a flash talk (2 min) to present AiiDA and the work on AURORA.

Xing:
* Had a discussion with Louis Ponet about the declarative workchain and nodetree programming.
* Finished first working versoin AiidaNodetree workchain.

Kristjan:
* [Materials Cloud React header component]( https://github.com/materialscloud-org/react-materialscloud-header)
* updated
* https://acwf-verification.materialscloud.org/
* https://mc3d.materialscloud.org/

Giovanni:
* MARVEL report: thanks!
* MARVEL retreat: everybody registered? Who is not coming? Marnik and Simon will present. The rest, I will summarize. Good to check slides with Marnik 1 week before so I don't repeat content
* Roadmap: let's clean it up and assign realistic goals, possibly deprioritizing things if the time is not enough. Brainstorming meeting in early Jan? Maybe we meet in Grindelwald a bit earlier before dinner? Other ideas?
* Next meetings: Skip 28th Dec, meet on Wed 11th Jan

Sebastiaan:
* **Ensure that `ProcessNode.get_builder_restart` fully restores all inputs including `non_db` ones** [[#5801]](https://github.com/aiidateam/aiida-core/pull/5801): Refactored after initial discussion and now a new keyword `metadata` is added to the `InputPort`. All inputs that are not linked as a `Data` node but stored directly on the `ProcessNode` somehow need to be marked with `metadata=True`. Two questions:
* 1) is name `metadata` ok? Or maybe `is_metadata` as suggested by Leopold
* 2) Currently only used by `metadata` inputs defined by `Process` and `CalcJob` base classes in `aiida-core`. Should we encourage or even allow plugins to use this as well, or restrict to internal use for now
* **`ProcessFunction`: Add support for variadic arguments** [[#5691]](https://github.com/aiidateam/aiida-core/pull/5691): Addressed the comments from Giovanni. I would merge this since the alternative implementation is likely not going to happen anytime soon.
* **Add `verdi calcjob remotecat` for printing remote file of running calcjobs** [[#4861]](https://github.com/aiidateam/aiida-core/pull/4861): Opened by Bonan long time ago and was waiting for final review. I have rebased the branch, addressed merge conflicts and added tests. One open question: it has the `--monitor` option (suggested by Marnik) which is useful, but to do it, he adds functionality to `Transport.gotocomputer_command` to run commands on the remote. It is purely additive and not breaking anything, but I am wondering if this is the right place for it. More logical would be `Transport.exec_command_wait` except that this currently doesn't work since it waits for the command to finish, which is not the case with `tail -f`.
* **RFC: permit class member functions as calcfunctions** [[#4963]](https://github.com/aiidateam/aiida-core/pull/4963): Old PR by Tiziano. I have updated it with tests and think it is ready. Just one open question since although it now enables support for process functions as class members, inline defined process functions will no longer be valid cache sources. I would vote to merge this PR though.

### Specific Agenda Items