Demote x86_64-apple-darwin from Tier 1 to Tier 2 with host tools
#3841
Conversation
shepmaster
added
T-compiler
shepmaster
force-pushed
the
demote-x86_64-apple-darwin
branch
from
74515b9 to
84fcc97
Compare
shepmaster
force-pushed
the
demote-x86_64-apple-darwin
branch
from
84fcc97 to
242cfad
Compare
|
Starting FCP already, we don't have many options here. |
|
Team member @Noratrieb has proposed to merge this. The next step is review by the rest of the tagged team members:
Concerns:
Once a majority of reviewers approve (and at most 2 approvals are outstanding), this will enter its final comment period. If you spot a major issue that hasn't been raised at any point in this process, please speak up! See this document for info about what commands tagged team members can give me. |
rfcbot
added
proposed-final-comment-period
|
|
||
| The first release after this RFC is merged will be the last one with Tier 1 support for the `x86_64-apple-darwin` target. | ||
| The release after that will demote the target to Tier 2 with host tools, | ||
| which means it will not be tested by CI. |
There was a problem hiding this comment.
FWIW, we do run some tests for non-tier-1-targets, e.g. some i586 targets. We could probably do the same for x86_64-apple-darwin? IOW, the consequence of this RFC is that we stop promising that we run all the tests, and we'll drop what is impractical, but we don't have to drop everything immediately. We could, for instance, keep running the library tests under rosetta, which shouldn't take too long.
There was a problem hiding this comment.
Yes, it may be feasible for rust-lang/rust to run a subset of tests still for x86_64-apple-darwin, this RFCs just drops that promise. This wording is somewhat accurate in that it's definitely a possibility for us to not run any tests (as per Tier 2 support level), even if that seems unlikely.
There was a problem hiding this comment.
I think that would be good to communicate clearly, also in the blog post: we don't actually plan to drop all tests immediately, we'll continue testing some things under Rosetta where feasible, etc.
Or is our plan really to drop all tests entirely when we stop using the free macos-13 runners?
There was a problem hiding this comment.
Or is our plan really to drop all tests entirely when we stop using the free macos-13 runners?
I think the intention is to keep that open as a possibility in case it becomes necessary.
There was a problem hiding this comment.
Right, that's what I figured. I think this needs to be quite explicit, or people will conclude that we are dropping all tests immediately.
There was a problem hiding this comment.
I don't think we can necessarily guarantee that
The RFC currently basically suggests that we guarantee that we don't run any tests.
That is not the same as not guaranteeing that we run all the tests.
There was a problem hiding this comment.
The RFC currently basically suggests that we guarantee that we don't run any tests. That is not the same as not guaranteeing that we run all the tests.
So something like
This means that it is no longer guaranteed that the
x86_64-apple-darwintarget will be tested in CI
Which includes the possibility that no tests for x86_64-apple-darwin is run at all?
There was a problem hiding this comment.
Yes. It's definitely a possibility, but it's a matter of continuous evaluation.
There was a problem hiding this comment.
Is this better?
diff --git a/text/3841-demote-x86_64-apple-darwin.md b/text/3841-demote-x86_64-apple-darwin.md
index e1e72361..942446ae 100644
--- a/text/3841-demote-x86_64-apple-darwin.md
+++ b/text/3841-demote-x86_64-apple-darwin.md
@@ -71,7 +71,7 @@ we can see that `x86_64-apple-darwin` has substantially fewer downloads than `aa
The first release after this RFC is merged will be the last one with Tier 1 support for the `x86_64-apple-darwin` target.
The release after that will demote the target to Tier 2 with host tools,
-which means it will not be tested by CI.
+which means we no longer guarantee that it will be tested by CI.
Once this RFC is merged,
a blog post will be published on the main Rust Blog announcing the change to alert users of the demotion.
|
Friendly nudge for the remaining FCP reviewers (would not have done so if it's not as time-sensitive), as on 2025-09-01 GitHub will discontinue providing free macOS x86_64 runners for public repositories, which won't leave us a lot of time for doing the demotion accounting for FCP time. |
|
@rfcbot concern dont-commit-to-dropping-all-test-coverage (Not sure if I can raise concerns here? I'm in t-compiler but not a "maintainer".) I think we shouldn't prescribe in the RFC that we stop all tests. I think we should just say we stop guaranteeing that all tests are run. That's a big difference. The stats show there's still a lot of x86_64-apple users out there, so it's too early to drop all test coverage entirely. |
|
@rfcbot concern test-coverage-wording
|
Rendered