fix: Skip problematic cyclic dev-dependencies #16871
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
rustbot
added
the
S-waiting-on-review
Veykril
changed the title
|
@bors r+ |
|
☀️ Test successful - checks-actions |
|
Well, this obviously didn't work out as I expected. Not sure why I thought this would be a good solution, one does quite often crate dev-only dependencies in their own workspace (r-a is an example of that). Guess we'll need to be a bit smarter about this after all. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Implements a workaround for #14167, notably it does not implement the ideas surfaced in the issue, but takes a simpler to implement approach (and one that is more consistent).
Effectively, all this does is discard dev-dependency edges that go from a workspace library target to another workspace library target. This means, using a dev-dependency to another workspace member inside unit tests will always fail to resolve for r-a now, (instead of being order dependent and causing problems elsewhere) while things will work out fine in integration tests, benches, examples etc. This effectively acknowledges package cycles to be okay, but crate graph cycles to be invalid:
Quoting #14167 (comment)
We disallow the latter here, while continuing to support the former
(What's missing is to supress diagnostics for such unit tests, though not doing so might be a good deterrent, making devs avoid the pattern altogether)