Support uploading neuroglancer precomputed and N5

[frcroth]

URL of deployed dev instance (used for testing):

• https://uploadmoredatatypes.webknossos.xyz

Steps to test:

• Upload these datasets: https://drive.google.com/drive/folders/1Yp7FRQ0_sG8Fetd-MwCRjaXmaDMkLw5M?usp=drive_link

TODOs:

• Precomputed Multilayer
• N5

Issues:

• fixes Support uploading n5 and precomputed datasets #7544

---

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

• Updated changelog
• Updated documentation if applicable
• Removed dev-only changes like prints and application.conf edits
• Considered common edge cases
• Needs datastore update after deployment

[fm3]

Looking good :)

Do you think it would help to introduce a trait LayerWithMagLocators that implements a withMags method (and does the switch case over the different implementations there?). That way we wouldn’t have to do all of this switch-casing everywhere in the business logic? Did not test if this would work but maybe it’s worth trying.

Also, thanks for going through the docs! 👏

[fm3]

what is this doing? 🤔

Also, should probably be extracted to makeMagPathRelative since this is used twice here.

For N5/precomputed this is not needed?

[frcroth]

Is used for zarr because of the different directory format (for ngff and zarr, there is an additional directory e.g. color)

[frcroth]

Looking good :)

Do you think it would help to introduce a trait LayerWithMagLocators that implements a withMags method (and does the switch case over the different implementations there?). That way we wouldn’t have to do all of this switch-casing everywhere in the business logic? Did not test if this would work but maybe it’s worth trying.

Have a look now. I think it does make sense, but I am not sure I like the name. Is there something more abstract that seperates layers with maglocators from others?

[fm3]

Have a look now. I think it does make sense, but I am not sure I like the name. Is there something more abstract that seperates layers with maglocators from others?

Good question! I’m afraid, not at this point. Layers without magLocators are currently

• annotation layers
• dataset layers with wkw format

In fact, I think we should add magLocator support for wkw layers as well, either with a migration of the jsons or an adapter that can read WKWResolution objects and turn them into MagLocators. This would also enable remote wkw datasets. But that does not have a high priority atm and it should not block this PR.

[fm3]

🎉