Improve Zarr Import UI (and general editing of datasets)

[philippotto]

URL of deployed dev instance (used for testing):

• https://___.webknossos.xyz

Steps to test:

• In general do some monkeytesting with the following:
  • use new "Add Zarr UI" (e.g., with layer url == https://data-humerus.webknossos.org/data/zarr/scalable_minds/l4dense_motta_et_al_demo/segmentation)
  • edit an existing dataset (zarr or wkw)

Issues:

• fixes Improve datasource editing for remote zarr import #6409

---

(Please delete unneeded items, merge only when none are left open)

• Updated (unreleased) changelog
• Ready for review

[philippotto]

[ ] The default name for dataset is the same as the layer name of the first layer. I am not sure if this makes sense. In case of your test DS above, the dataset name should have been l4_dense_motta. I am not sure if the name can also be derived from the Zarr file/spec or with should either leave the field empty or with a generic name.

@fm3 I think the initial name is coming from the back-end, right? defaulting to an empty string sounds like a good solution for now (definitely, better than a poor guess, I'd say).

[philippotto]

• Maybe not technically caused by this PR but wK offers a config suggestion, i.e. to remove all datalayer (dataLayer = []) when clearly the advanced view still has content/values for this property...

@fm3 I fixed this in the front-end now. I think it's fair to say that the backend only infers what's on disk. So, adding existing zarr layers to the inferred datasource makes (some) sense to me. Let me know in case you disagree 🤙

[fm3]

backend only infers what's on disk

Yes, makes sense to me 👍

initial name is coming from the back-end, right?

correct

defaulting to an empty string sounds like a good solution for now (definitely, better than a poor guess, I'd say).

ok I’ll include that change in #6520 to avoid merge conflicts, that pr should be merged soon

[philippotto]

@hotzenklotz Thanks for the thorough testing. I debugged and integrated your ideas as best as possible. Please test again (the new dev instance is already deployed).

Another nitpick: There is no easy way to add layers to an existing dataset later. Yes, I can manually add a layer to the datasource through the advanced JSON field but that is error prone. (I tried adding a Zarr color layer, which required me to remember the "magic"/"hidden" properties, "dataFormat": "zarr". The URL was also incorrect in my case

This is the only thing I didn't do. The form handling with antd is quite error-prone. I'd defer this feature to another iteration (maybe in combination to when the dataset management topic is tackled) to not drag this PR any longer.

[hotzenklotz]

I re-tested and LGTM. Ready to merge :-)

[philippotto]

I re-tested and LGTM. Ready to merge :-)

I need your official acceptance to merge this PR :)