You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The code formatting action appears to be broken, and no longer reports incorrect formatting. See e.g. #97029, but I've noticed this in multiple PRs recently. Not sure when exactly it broke.
The code formatting action appears to be broken, and no longer reports incorrect formatting. See e.g. https://github.com//pull/97029, but I've noticed this in multiple PRs recently. Not sure when exactly it broke.
It doesn't even seem like the job is running currently? There are quite a few other jobs missing in new PRs as well. The last code formatting job currently ran 25+ minutes ago and PRs have definitely been updated more recently.
I also can't get the job to run on my fork currently.
It seems like maybe just pull_request jobs? All the pull_request_target jobs, like the PR labelers, seem to be working fine.
Do you mean the case where a single libc++ header is modified and that header has no extension?
A single c++ file in llvm/. I haven't done a lot of investigation, but something like the following diff would pass even though clang format should fail it:
diff --git a/llvm/lib/CodeGen/MLRegAllocEvictAdvisor.cpp b/llvm/lib/CodeGen/MLRegAllocEvictAdvisor.cpp
index 5698f6d6fea00..2c3e0f782c567 100644
--- a/llvm/lib/CodeGen/MLRegAllocEvictAdvisor.cpp+++ b/llvm/lib/CodeGen/MLRegAllocEvictAdvisor.cpp@@ -10,6 +10,8 @@
//
//===----------------------------------------------------------------------===//
+// This line should never get through the formatting job clang-format please format this into something that is actually reasonable.+
#include "AllocationOrder.h"
#include "RegAllocEvictionAdvisor.h"
#include "RegAllocGreedy.h"
Activity
llvmbot commentedon Jun 28, 2024
@llvm/issue-subscribers-infrastructure
Author: Nikita Popov (nikic)
tstellar commentedon Jun 28, 2024
Maybe this one: 5914a56 ?
It's not just that the comment is missing, the job is passing even if there is incorrect formatting, right?
nikic commentedon Jun 28, 2024
Yeah, the job is passing.
7620fe0 is another possible culprit.
boomanaiden154 commentedon Jun 28, 2024
I'll take a look.
boomanaiden154 commentedon Jun 28, 2024
It doesn't even seem like the job is running currently? There are quite a few other jobs missing in new PRs as well. The last code formatting job currently ran 25+ minutes ago and PRs have definitely been updated more recently.
I also can't get the job to run on my fork currently.
It seems like maybe just
pull_requestjobs? All thepull_request_targetjobs, like the PR labelers, seem to be working fine.boomanaiden154 commentedon Jun 28, 2024
https://www.githubstatus.com/incidents/9vwllhs2w1kj
I guess I'll wait for that to be fixed and then get to investigating.
tstellar commentedon Jun 28, 2024
Here is a job from a few hours ago that correctly failed: https://github.com/llvm/llvm-project/actions/runs/9714796529
boomanaiden154 commentedon Jun 28, 2024
It correctly failed on my fork too: https://github.com/boomanaiden154/llvm-project/actions/runs/9716540595/job/26820327869?pr=48
I wonder if it's the one cpp file case?
boomanaiden154 commentedon Jun 28, 2024
It's 7620fe0.
It breaks at least the one cpp file case.
I'll revert to get things back to the way they were and ping on the PR.
Revert "[CI][format] Explicitly pass extensions to git-clang-format (#…
Revert "[CI][format] Explicitly pass extensions to git-clang-format (l…
ldionne commentedon Jul 4, 2024
Do you mean the case where a single libc++ header is modified and that header has no extension?
Thanks, I'll take a look. I was away at a WG21 meeting last week.
boomanaiden154 commentedon Jul 4, 2024
A single c++ file in
llvm/. I haven't done a lot of investigation, but something like the following diff would pass even though clang format should fail it:ldionne commentedon Jul 9, 2024
@boomanaiden154 Thanks, investigating in #98227
ldionne commentedon Aug 29, 2024
This should have been closed when I merged #98227.