[vim/vim] CI: two mingw Windows jobs time out since 2026-08-05 (Issue #20976)

h_east (Vim Github Repository) <[email protected]> Sat, 08 Aug 2026 07:21:54 -0700
Newsgroups gmane.editors.vim.devel
Message-ID <vim/vim/issues/[email protected]>
h-east created an issue (vim/vim#20976)

Since 2026-08-05 the two `HUGE, mingw, VIMDLL:yes` Windows jobs fail on every
master run:

    The background step 'Test and show the result of testing gVim' has timed
    out after 15 minutes.

They are the only jobs that run the gVim and the Vim test suites at the same
time on a HUGE build: the step conditions are `GUI == 'yes' || VIMDLL == 'yes'`
and `GUI == 'no' || VIMDLL == 'yes'`, so `VIMDLL: yes` runs both. The other two
`VIMDLL: yes` jobs are NORMAL and TINY and pass.

Two separate things are going on.

**1. The parallel test steps.** Reverting c0fc51184 ("CI: MS-Windows: Run
gvim/vim tests in parallel") makes the x64 job pass. Without it the gVim tests
reach `ALL DONE` with `Failed: 0 Tests` and then hang for eight minutes with no
output until the step is killed, after which the runner reports:

    Terminate orphan process: pid (748) (gvim)
    Terminate orphan process: pid (972) (vimrun)
    Terminate orphan process: pid (1256) (conhost)

The tests pass; a leftover child keeps the step from finishing.

**2. `test_assert.vim` stalls on x86.** With the parallel change reverted, x86
still fails, but the run stops inside `test_assert.vim` and never reaches
`ALL DONE`. A probe listing leftover `gvim.exe` after every test file shows
nothing is left over before it:

    ==== after test_arglist.vim
    INFO: No tasks are running which match the specified criteria.
    ..\gvim ... -S runtest.vim test_assert.vim
    NMAKE : fatal error U1058: terminated by user

`test_assert.vim` is the only file in the suite that calls `RunVim()`, in
`Test_assert_equal_typed()`.

No Vim change is involved: building 5a1124e4f, the last commit whose mingw jobs
were all green, and running it today reproduces both failures exactly. The
workflow files have not changed either since the dependabot bump on 2026-07-25.

I could not reproduce any of it locally with an MSVC x64 build, neither the
whole `test_assert.vim` nor `Test_assert_equal_typed()` alone nor the same
command by hand through `:silent !`, so something about the CI environment is
needed to trigger it.

I can send a PR reverting the parallel steps, which fixes the x64 job. For x86
I would need to narrow it down with `TEST_FILTER` on CI itself, unless someone
recognises the symptom.

-- 
Reply to this email directly or view it on GitHub:
https://github.com/vim/vim/issues/20976
You are receiving this because you are subscribed to this thread.

Message ID: <vim/vim/issues/[email protected]>

-- 
-- 
You received this message from the "vim_dev" maillist.
Do not top-post! Type your reply below the text you are replying to.
For more information, visit http://www.vim.org/maillist.php

--- 
You received this message because you are subscribed to the Google Groups "vim_dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
To view this discussion visit https://groups.google.com/d/msgid/vim_dev/vim/vim/issues/20976%40github.com.