Feedback for the Upload-Portal

[MagnusSletten]

The upload portal can be found here: https://upload-portal.nmrlipids.fi/

If you have any thoughts on how it could be improved or if you’ve encountered any problems, please don’t hesitate to share them here.

[comcon1]

I have finished addition of my second system. Everything was as expected. There are issues regarding visualization and interactions with workflows, but these issues I have already added. The most important is to have pre-analysis as well as AddData dry-run, it will save a lot of time by just reporting problems. Some typical errors like atomnames in mapping files could be immidiately caught on AddData dry-run stage. The corresponding issues I have opened.

[MagnusSletten]

That is what my issue was about. NMRLipids/FAIRMD_lipids#150 I mentioned it. If mapping file has wrong naming, we should check it during AddFile dry-run. Currently it fails only on the analysis phase. AddFile now checks only number of atoms.

I wonder if we should separate the dry run in its own action (right now it's a step in the addition-workflow) which we allow to run immediately upon creating the PR. The action could just report any errors directly to the PR itself. The best solution, which probably takes too much work right now, is to have a very lightweight check for mapping files which the webserver could do itself but the separation of dry run is maybe a decent alternative.

There are issues regarding visualization and interactions with workflows, but these issues I have already added.

Where are the workflow issues?

[comcon1]

I wonder if we should separate the dry run in its own action (right now it's a step in the addition-workflow) which we allow to run immediately upon creating the PR.

Yes, that is a more "pipeline-ish" way. Can you please create an issue about it?

Where are the workflow issues?

I think I thought about the form complains which you have already done or are working on, removing some fields, separate lipid and solvent, add combo-boxes. That kind of things. So, don't worry about this comment, it's already in play.

[MagnusSletten]

@ohsOllila, it should be ready for testing now.

There's now a separate workflow which runs immediately after creating a PR that runs dry run of addData to check mapping files and rule out other errors. There shouldn't be a problem with memory either but this will be improved further when we move to 16GB runner which I'll do this week.