Switch to root folder if current folder cannot be loaded

[philippotto]

URL of deployed dev instance (used for testing):

• https://___.webknossos.xyz

Steps to test:

• create a new folder
• yarn refresh-schema
• reload page

Issues:

• fixes When dashboard localstorage points to deleted folder, go to root instead #7658

---

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

• Updated changelog

[fm3]

I’m afraid this doesn’t quite work for me yet 😬 It looks like some folders briefly flash but then it switches to the blank version again after a few ms. Does it work for you, did you test with refresh-schema? I don’t know what’s different, maybe it’s a firefox vs chrome thing? 🤔

[philippotto]

I’m afraid this doesn’t quite work for me yet 😬 It looks like some folders briefly flash but then it switches to the blank version again after a few ms. Does it work for you, did you test with refresh-schema? I don’t know what’s different, maybe it’s a firefox vs chrome thing? 🤔

Initially, I simply tried localStorage.setItem("sample@scm.io-sample_organization-activeFolderId", "65e871d4ec0000dc04353aee") and then reloaded the page. That worked even on master. Then, I tried yarn refresh-schema which was problematic. With my fix, a simple reload worked. However, something might have changed inbetween.
Maybe the folder hierarchy is cached internally and if that cache is outdated, the switch to the root doesn't work.
I'll look into it again 🤷

It's a bit annoying to test since refreshing the schema takes so long. Do you know whether there is a faster way to clear the folders?

[fm3]

I’m afraid manually editing folders in the DB is not going to be simpler or faster than yarn refresh-schema :/

[philippotto]

Please try again! I think it already worked before, but there was quite a bit of a delay until the switch would happen. The active folder is fetched and if that fails, it is retried three times. Due to an exponential back off in the delay time (default of react-query), this took at least 6 seconds. Now, it's down to 1.5s. I hope this wait is acceptable :)

By the way, deleting the rm -f target/scala-2.13/src_managed/schema/com/scalableminds/webknossos/schema/Tables.scala bit from yarn refresh-schema helped a bit to get feedback faster :)

[fm3]

Yes, seems to work now, this is great! One more tiny thing: could the error toast saying »could not find the requested folder« also vanish after some seconds? If that would be simple, I’d appreciate it, otherwise this change as it is also helps me a lot already :)

[philippotto]

Unfortunately, it's not trivial :S