[Chapel GA Tests] 2026-09-14

[chapelhpcBot]

The following GitHub Actions jobs failed:

• extended-checks / test-homebrew (ubuntu-latest), step: "run homebrew testing"
• extended-checks / test-homebrew (macos-latest), step: "run homebrew testing"
• full-checks / Chapel paratest on linux64 with C backend / Build Chapel (linux64-c-backend), step: "build Chapel"
• full-checks / Chapel paratest on linux64 with C backend / Aggregate results (linux64-c-backend), step: "aggregate and report results"
• full-checks / Chapel paratest on darwin / Test test/chplvis (darwin), step: "download Chapel build"
• full-checks / Chapel paratest on darwin / Aggregate results (darwin), step: "aggregate and report results"

Ran 593 total jobs at commit 9b839e1b6bc121f02c5daad801ead50136b9d193 "Update prereqs that highest supported LLVM is 22 (#29398)".

GitHub Actions run info:

• SHA: 9b839e1b6bc121f02c5daad801ead50136b9d193
• Run ID: 34819295338
• Run attempt: 1
• Run number: 23
• Triggering event name: schedule

[benharsh]

The C backend failures are related to each other, due to the "build" step failing to complete as a result of a network failure.

The chplvis test failed to download the Chapel artifact, except it was due to a "403 Forbidden" error. @annarift , have we seen this before?

In the aggregate results we see the following, which doesn't immediately make sense to me:

Failed test jobs: 0
Overall result: FAIL

Maybe we need some additional debugging output?

[annarift]

@benharsh

The chplvis test failed to download the Chapel artifact, except it was due to a "403 Forbidden" error. @annarift , have we seen this before?

I don't think we've seen it before. I don't think it's a correct error from Github, as if there was truly a permissions issue then all the other sibling jobs that download the build artifact should be hitting it as well. I suspect it's a Github blip that will go away next run.

@benharsh

In the aggregate results we see the following, which doesn't immediately make sense to me:

Failed test jobs: 0
Overall result: FAIL

Maybe we need some additional debugging output?

I figure the weird output from the c-backend aggregation job is because the c-backend build job failed, so there were no test runs to aggregate results from. I think we ought to either not run the aggregation job or have it give better output in this case, I'll look into that.

[annarift]

PR Report a build failure up front in paratest results aggregation by arifthpe · Pull Request #29403 · chapel-lang/chapel · GitHub makes the aggregation job just report that the build failed in that case.