Fallback on non-MavenRepository repos

[kubukoz]

Currently, if you publish to the new Sonatype Central repository (#721), during publishing you'll encounter the following issue:

[error] scala.MatchError: URLRepository(https://settingkey/(This / This / This / sonatypeCredentialHost), Patterns(ivyPatterns=Vector(), artifactPatterns=Vector(), isMavenCompatible=false, descriptorOptional=false, skipConsistencyCheck=false), false) (of class sbt.librarymanagement.URLRepository)

It happens in the code that computes a step summary. Originally, I tried to write a rendering function for the entire URLRepository, but serializing Patterns is a bit tedious and the step summary isn't all that useful, so I figured we can start with an empty string for now.

[aartigao]

serializing Patterns is a bit tedious and the step summary isn't all that useful

While this is true, I think returning at least URLRepository.name will be better in this case. And I'll keep the empty string for global fallback.

[kubukoz]

yo @armanbilge can we get this merged&released please? 🥺

[armanbilge]

I think returning at least URLRepository.name will be better in this case.

I agree.

the step summary isn't all that useful

I find it very useful :) I copy-paste from there to get the resolver config whenever I need to depend on a snapshot.

The goal is to print something there that will help the user.

[armanbilge]

Btw, thanks so much for investigating this, very appreciated 🙏

[kubukoz]

Fair enough, added more verbosity! (in non-URLRepository just doing a full toString as well)

[kubukoz]

TBH I don't think that'll fix it.

URLRepository(https://SettingKey(This / This / This / sonatypeCredentialHost)

this would imply that there was a toString on a SettingKey used instead of a .value (or equivalent) :/

[aartigao]

TBH I don't think that'll fix it.

WDYM? At least the MatchError will be fixed, right? You mean that after fixing that it's still failing?

[kubukoz]

TBH I don't think that'll fix it.

WDYM? At least the MatchError will be fixed, right? You mean that after fixing that it's still failing?

I haven't tried out the change yet, just thinking that if the URLRepository is constructed this way, and we use that to publish, then it's likely wrong.

But maybe it's not actually used for publishing and only used to print the result, in which case it'd be ok 😅

[armanbilge]

thanks for chasing this down!