Any hints on fixing iOS builds in a PR? <https://g...

# confetti

[yschimke]

Any hints on fixing iOS builds in a PR? https://github.com/joreilly/Confetti/pull/549

[John O'Reilly]

looks like iOS test timed out

[John O'Reilly]

Test case 'ConfettiTests.testGetSessions()' failed on 'Clone 1 of iPhone 8 - Confetti (17837)' (60.000 seconds)
Error: Process completed with exit code 65.

[John O'Reilly]

we have 60s timeout

[John O'Reilly]

was there any changes made to query code?

[John O'Reilly]

this is the tests it's running https://github.com/joreilly/Confetti/blob/main/iosApp/ConfettiTests/ConfettiTests.swift

[mbonnin]

It's a shame stdout is not displayed during tests

[mbonnin]

Trying to run the test locally here

[John O'Reilly]

hmm, true

[John O'Reilly]

maybe there's something we can do to influence that

[mbonnin]

Looks like it fails with

com.apollographql.apollo3.exception.CacheMissException: Object 'QUERY_ROOT' has no field named 'bookmarks'

[mbonnin]

But maybe that's just a log, not the actual error

[mbonnin]

Alright, I think because of the bookmarksConnection thing, it creates a cache miss because the connection deosn't have an id

[mbonnin]

So this was actually a good test

[mbonnin]

Looking into it

[mbonnin]

Weird that it's looking for "bookmark" though

[mbonnin]

Well yea never mind, not related

[mbonnin]

Ah yea, got it now. toFlow().first() is different from execute()

[mbonnin]

execute() filters out the errors

[mbonnin]

so that it returns a single response

[mbonnin]

toFlow emits everything

[mbonnin]

Hopefully fixed with https://github.com/joreilly/Confetti/pull/549/commits/ee7fb291df042f7cbf654fe5ad30ddb9b9503109

[mbonnin]

Need to step away from keyboard for a bit. Will think if we can do something better

[yschimke]

Thanks, both.

[mbonnin]

Thank you for surfacing this. This is def a footgun of CacheAndNetwork. Will think of improving this for v4