[PATCH 0/7] Some assorted fixes for GitLab CI

[PATCH 0/7] Some assorted fixes for GitLab CI

[Patrick Steinhardt]

Hi,

I recently had the pleasure of debugging a couple of failing
MSVC+Windows jobs in GitLab CI, which hasn't been quite fun because we
didn't know to print error logs, and neither did we upload the failed
test artifacts. This patch series is the result of this frustration and
fixes a couple of smaller issues in the context of our CI:

  - I noticed that test slicing is slightly wrong because of a
    difference between zero- and one-based indices, which causes us to
    skip the first test on GitLab.

  - I deduplicated how we run Meson tests so that both GitLab and GitHub
    use the same "run-test-slice-meson.sh" script.

  - I add logic to handle failing tests via "print-test-failures.sh".

The result can be found at [1]. Note that tests are failing, but those
failures are fixed in a separate patch series via [2]. In any case, I
guess those test failures also serve as a good demonstration how the
failing tests show up now.

Changes in v2:
  - Fix syntax in GitHub workflow.
  - Add another patch on top that fixes win+Meson tests in GitHub. Those
    tests didn't execute at all, even before this patch series.
  - This uncovers another bug in how we set up MERGE_TOOLS_DIR, so this
    results in another commit.
  - I've now tested also tested the changes on GitHub [3].
  - Link to v1: https://lore.kernel.org/r/20260209-b4-pks-ci-msvc-iconv-fixes-v1-0-1e3167cd8828@pks.im

Thanks!

Patrick

[1]: https://gitlab.com/gitlab-org/git/-/merge_requests/497
[2]: <20260209-b4-pks-ci-msvc-iconv-fixes-v1-0-1e3167cd8828@pks.im>
[3]: https://github.com/git/git/pull/2195

---
Patrick Steinhardt (7):
      ci: handle failures of test-slice helper
      ci: don't skip smallest test slice in GitLab
      meson: fix MERGE_TOOL_DIR with "--no-bin-wrappers"
      github: fix Meson tests not executing at all
      ci: make test slicing consistent across Meson/Make
      gitlab-ci: use "run-test-slice-meson.sh"
      gitlab-ci: handle failed tests on MSVC+Meson job

 .github/workflows/main.yml |  6 +++---
 .gitlab-ci.yml             | 17 +++++++++++++++--
 ci/run-test-slice-meson.sh |  2 +-
 ci/run-test-slice.sh       |  6 +++---
 t/helper/test-path-utils.c | 18 ++++++++++++------
 t/meson.build              |  1 +
 6 files changed, 35 insertions(+), 15 deletions(-)


---
base-commit: 3e0db84c88c57e70ac8be8c196dfa92c5d656fbc
change-id: 20260209-b4-pks-ci-meson-improvements-93d8a1ffdd27

[PATCH 1/7] ci: handle failures of test-slice helper

[Patrick Steinhardt]

The "run-test-slice.sh" script executes the test helper to slice up
tests passed to it. As the execution is part of a pipe though, we end up
ignoring any potential error code returned by the helper.

Make the code more robust by storing the tests in a variable first so
that we can split up the pipeline.

Signed-off-by: Patrick Steinhardt <ps@pks.im>
---
 ci/run-test-slice.sh | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/ci/run-test-slice.sh b/ci/run-test-slice.sh
index 0444c79c02..ff948e397f 100755
--- a/ci/run-test-slice.sh
+++ b/ci/run-test-slice.sh
@@ -5,9 +5,9 @@
 
 . ${0%/*}/lib.sh
 
-group "Run tests" make --quiet -C t T="$(cd t &&
-	./helper/test-tool path-utils slice-tests "$1" "$2" t[0-9]*.sh |
-	tr '\n' ' ')" ||
+TESTS=$(cd t && ./helper/test-tool path-utils slice-tests "$1" "$2" t[0-9]*.sh)
+
+group "Run tests" make --quiet -C t T="$(echo "$TESTS" | tr '\n' ' ')" ||
 handle_failed_tests
 
 # We only have one unit test at the moment, so run it in the first slice

-- 
2.53.0.295.g64333814d3.dirty

[PATCH 2/7] ci: don't skip smallest test slice in GitLab

[Patrick Steinhardt]

The "ci/run-test-slice.sh" script can be used to slice up all of our
tests into N pieces and then run each of them on a separate CI job.
This is used by both GitLab and GitHub CI to speed up Windows tests,
which would otherwise be painfully slow.

The infra itself is fueled by `test-tool path-utils slice-tests`. This
tool receives as input an "offset" and a "stride" that can be combined
to slice up tests. This framing can be misleading though: you are
expected to pass a zero-based index as "offset", and the complete number
of slices to the "stride". The latter makes sense, but it is somewhat
surprising that the offset needs to be zero-based. And this is in fact
biting us: while GitHub passes zero-based indices, GitLab passes
`$CI_NODE_INDEX`, which is a one-based indice.

Ideally, we should have verification that the parameters make sense.
And naturally, one would for example expect that it's an error to call
the binary with an offset larger than the stride. But with the current
framing as "offset" it's not even wrong to do so, as it is of course
well-defined to start at a larger offset than the stride.

This means that we get this wrong on GitLab's CI, as we pass a one based
index there, and this causes us to skip one of the tests. Interestingly,
it's not the lexicographically first test that we skip. Instead, as we
sort tests by size before slicing them, we skip the _smallest_ test.

Reframe the problem to instead talk about "slice number" and "total
number of slices". For all of our use cases this is semantically
equivalent, but it allows us to perform some verifications:

  - The total number of slices must be greater than 1.

  - The selected slice must be between 1 <= nr <= slices_total.

As the indices are now one-based it means that GitLab's CI is fixed.
The GitHub workflow is updated accordingly.

Signed-off-by: Patrick Steinhardt <ps@pks.im>
---
 .github/workflows/main.yml |  2 +-
 t/helper/test-path-utils.c | 18 ++++++++++++------
 2 files changed, 13 insertions(+), 7 deletions(-)

diff --git a/.github/workflows/main.yml b/.github/workflows/main.yml
index f2e93f5461..ebb1a06bfa 100644
--- a/.github/workflows/main.yml
+++ b/.github/workflows/main.yml
@@ -150,7 +150,7 @@ jobs:
     - uses: git-for-windows/setup-git-for-windows-sdk@v1
     - name: test
       shell: bash
-      run: . /etc/profile && ci/run-test-slice.sh ${{matrix.nr}} 10
+      run: . /etc/profile && ci/run-test-slice.sh $((${{matrix.nr}} + 1)) 10
     - name: print test failures
       if: failure() && env.FAILED_TEST_ARTIFACTS != ''
       shell: bash
diff --git a/t/helper/test-path-utils.c b/t/helper/test-path-utils.c
index f5f33751da..874542ec34 100644
--- a/t/helper/test-path-utils.c
+++ b/t/helper/test-path-utils.c
@@ -477,14 +477,20 @@ int cmd__path_utils(int argc, const char **argv)
 
 	if (argc > 5 && !strcmp(argv[1], "slice-tests")) {
 		int res = 0;
-		long offset, stride, i;
+		long slice, slices_total, i;
 		struct string_list list = STRING_LIST_INIT_NODUP;
 		struct stat st;
 
-		offset = strtol(argv[2], NULL, 10);
-		stride = strtol(argv[3], NULL, 10);
-		if (stride < 1)
-			stride = 1;
+		slices_total = strtol(argv[3], NULL, 10);
+		if (slices_total < 1)
+			die("there must be at least one slice, got '%s'",
+			    argv[3]);
+
+		slice = strtol(argv[2], NULL, 10);
+		if (1 > slice || slice > slices_total)
+			die("slice must be in the range 1 <= slice <= %ld, got '%s'",
+			    slices_total, argv[2]);
+
 		for (i = 4; i < argc; i++)
 			if (stat(argv[i], &st))
 				res = error_errno("Cannot stat '%s'", argv[i]);
@@ -492,7 +498,7 @@ int cmd__path_utils(int argc, const char **argv)
 				string_list_append(&list, argv[i])->util =
 					(void *)(intptr_t)st.st_size;
 		QSORT(list.items, list.nr, cmp_by_st_size);
-		for (i = offset; i < list.nr; i+= stride)
+		for (i = slice - 1; i < list.nr; i+= slices_total)
 			printf("%s\n", list.items[i].string);
 
 		return !!res;

-- 
2.53.0.295.g64333814d3.dirty

[PATCH 3/7] meson: fix MERGE_TOOL_DIR with "--no-bin-wrappers"

[Patrick Steinhardt]

On Windows, we execute tests with "--no-bin-wrappers". This has been
introduced via a87e427e35 (ci: speed up Windows phase, 2019-01-29) to
save some time: spawning processes is expensive on Windows, and shell
scripts tend to spawn a bunch of them. So overall, the bin-wrappers led
to a performance overhead of ~10-30%.

This causes test failures when using Meson on Windows:

  failure: t7610.28 mergetool --tool-help shows recognized tools
    ++ git mergetool --tool-help
    /d/a/git/git/build/git-mergetool--lib: line 45: cd: D:/a/git/git/build/mergetools: No such file or directory

The root cause here is that our bin-wrappers are usually responsible for
setting up the `MERGE_TOOL_DIR` environment variable so that we can
locate these scripts. But as we don't use the bin-wrappers, we'll
instead use the default location for merge tools, which is derived from
`GIT_EXEC_PATH`. And as `GIT_EXEC_PATH` points to our build directory,
which won't ever contain any of the merge tools, we will fail to locate
any of the merge tools.

This issue has went unnoticed for a long time given that we only skip
bin-wrappers on Windows, and because the CI jobs on Windows didn't
execute due to a bug.

Fix the issue by always setting the `MERGE_TOOL_DIR` environment
variable to the correct directory.

Signed-off-by: Patrick Steinhardt <ps@pks.im>
---
 t/meson.build | 1 +
 1 file changed, 1 insertion(+)

diff --git a/t/meson.build b/t/meson.build
index 19e8306298..5ba7efd603 100644
--- a/t/meson.build
+++ b/t/meson.build
@@ -1208,6 +1208,7 @@ endif
 
 test_environment = script_environment
 test_environment.set('GIT_BUILD_DIR', git_build_dir)
+test_environment.set('MERGE_TOOLS_DIR', meson.project_source_root() / 'mergetools')
 
 foreach integration_test : integration_tests
   test(fs.stem(integration_test), shell,

-- 
2.53.0.295.g64333814d3.dirty

[PATCH 4/7] github: fix Meson tests not executing at all

[Patrick Steinhardt]

While the win+Meson test jobs run in GitHub workflows, the shell script
that is supposed to run the jobs is seemingly not running at all. All
that the CI job prints is the following:

  Run ci/run-test-slice-meson.sh build 1 10
    ci/run-test-slice-meson.sh build 1 10
    shell: C:\Program Files\PowerShell\7\pwsh.EXE -command ". '{0}'"
    env:
      DEVELOPER: 1

The step is currently defined to use PowerShell, and of course it
doesn't know how to execute POSIX shell scripts. What's surprising
though is that this step doesn't even lead to a CI failure.

Fix the issue by using Bash instead of PowerShell, as we do in other
steps that execute shell scripts.

Signed-off-by: Patrick Steinhardt <ps@pks.im>
---
 .github/workflows/main.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/workflows/main.yml b/.github/workflows/main.yml
index ebb1a06bfa..76ef513e72 100644
--- a/.github/workflows/main.yml
+++ b/.github/workflows/main.yml
@@ -297,7 +297,7 @@ jobs:
         name: windows-meson-artifacts
         path: build
     - name: Test
-      shell: pwsh
+      shell: bash
       run: ci/run-test-slice-meson.sh build ${{matrix.nr}} 10
     - name: print test failures
       if: failure() && env.FAILED_TEST_ARTIFACTS != ''

-- 
2.53.0.295.g64333814d3.dirty

[PATCH 5/7] ci: make test slicing consistent across Meson/Make

[Patrick Steinhardt]

In the preceding commit we have adjusted test slicing to be one-based
when using the "ci/run-test-slice.sh" script. But we also have an
equivalent script for Meson that is still zero-based, which is of course
inconsistent.

Adapt the script to be one-based, as well, and adapt the GitHub workflow
accordingly. Note that GitLab doesn't yet use the script, so it does not
need to be adapted. This will change in the next commit though.

Signed-off-by: Patrick Steinhardt <ps@pks.im>
---
 .github/workflows/main.yml | 2 +-
 ci/run-test-slice-meson.sh | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/.github/workflows/main.yml b/.github/workflows/main.yml
index 76ef513e72..352357965b 100644
--- a/.github/workflows/main.yml
+++ b/.github/workflows/main.yml
@@ -298,7 +298,7 @@ jobs:
         path: build
     - name: Test
       shell: bash
-      run: ci/run-test-slice-meson.sh build ${{matrix.nr}} 10
+      run: ci/run-test-slice-meson.sh build $((${{matrix.nr}} + 1)) 10
     - name: print test failures
       if: failure() && env.FAILED_TEST_ARTIFACTS != ''
       shell: bash
diff --git a/ci/run-test-slice-meson.sh b/ci/run-test-slice-meson.sh
index 961c94fba0..a6df927ba5 100755
--- a/ci/run-test-slice-meson.sh
+++ b/ci/run-test-slice-meson.sh
@@ -9,5 +9,5 @@
 
 group "Run tests" \
 	meson test -C "$1" --no-rebuild --print-errorlogs \
-		--test-args="$GIT_TEST_OPTS" --slice "$((1+$2))/$3" ||
+		--test-args="$GIT_TEST_OPTS" --slice "$(($2))/$3" ||
 handle_failed_tests

-- 
2.53.0.295.g64333814d3.dirty

Re: [PATCH 5/7] ci: make test slicing consistent across Meson/Make

[Johannes Schindelin]

Hi Patrick,

On Wed, 11 Feb 2026, Patrick Steinhardt wrote:

> In the preceding commit we have adjusted test slicing to be one-based
> when using the "ci/run-test-slice.sh" script. But we also have an
> equivalent script for Meson that is still zero-based, which is of course
> inconsistent.
> 
> Adapt the script to be one-based, as well, and adapt the GitHub workflow
> accordingly. Note that GitLab doesn't yet use the script, so it does not
> need to be adapted. This will change in the next commit though.

This seems to require the following to be squashed in:

-- snip --
From 677370c8cd3b73358713a0efac6bb573b347d53f Mon Sep 17 00:00:00 2001
From: Johannes Schindelin <johannes.schindelin@gmx.de>
Date: Wed, 18 Feb 2026 22:12:27 +0100
Subject: [PATCH] fixup??? ci: make test slicing consistent across Meson/Make

This seems to be needed to let the first "win+VS test" job pass. See
https://github.com/git-for-windows/git/actions/runs/22153194896/job/64053953314#step:5:70
("fatal: slice must be in the range 1 <= slice <= 10, got '0'").

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
---
 .github/workflows/main.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/workflows/main.yml b/.github/workflows/main.yml
index 412842eb3e2..42585fff779 100644
--- a/.github/workflows/main.yml
+++ b/.github/workflows/main.yml
@@ -240,7 +240,7 @@ jobs:
       shell: bash
       env:
         NO_SVN_TESTS: 1
-      run: . /etc/profile && ci/run-test-slice.sh ${{matrix.nr}} 10
+      run: . /etc/profile && ci/run-test-slice.sh $((${{matrix.nr}}+1)) 10
     - name: print test failures
       if: failure() && env.FAILED_TEST_ARTIFACTS != ''
       shell: bash
-- snap --

This here run proves that this change fixes the CI failure:
https://github.com/git-for-windows/git/actions/runs/22157880378/job/64067214717#step:5:70

Patrick, would you mind adopting this amendment into your patch?

Thank you,
Johannes

> 
> Signed-off-by: Patrick Steinhardt <ps@pks.im>
> ---
>  .github/workflows/main.yml | 2 +-
>  ci/run-test-slice-meson.sh | 2 +-
>  2 files changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/.github/workflows/main.yml b/.github/workflows/main.yml
> index 76ef513e72..352357965b 100644
> --- a/.github/workflows/main.yml
> +++ b/.github/workflows/main.yml
> @@ -298,7 +298,7 @@ jobs:
>          path: build
>      - name: Test
>        shell: bash
> -      run: ci/run-test-slice-meson.sh build ${{matrix.nr}} 10
> +      run: ci/run-test-slice-meson.sh build $((${{matrix.nr}} + 1)) 10
>      - name: print test failures
>        if: failure() && env.FAILED_TEST_ARTIFACTS != ''
>        shell: bash
> diff --git a/ci/run-test-slice-meson.sh b/ci/run-test-slice-meson.sh
> index 961c94fba0..a6df927ba5 100755
> --- a/ci/run-test-slice-meson.sh
> +++ b/ci/run-test-slice-meson.sh
> @@ -9,5 +9,5 @@
>  
>  group "Run tests" \
>  	meson test -C "$1" --no-rebuild --print-errorlogs \
> -		--test-args="$GIT_TEST_OPTS" --slice "$((1+$2))/$3" ||
> +		--test-args="$GIT_TEST_OPTS" --slice "$(($2))/$3" ||
>  handle_failed_tests
> 
> -- 
> 2.53.0.295.g64333814d3.dirty
> 
> 
> 

Re: [PATCH 5/7] ci: make test slicing consistent across Meson/Make

[Patrick Steinhardt]

On Wed, Feb 18, 2026 at 10:32:20PM +0100, Johannes Schindelin wrote:
> Hi Patrick,
> 
> On Wed, 11 Feb 2026, Patrick Steinhardt wrote:
> 
> > In the preceding commit we have adjusted test slicing to be one-based
> > when using the "ci/run-test-slice.sh" script. But we also have an
> > equivalent script for Meson that is still zero-based, which is of course
> > inconsistent.
> > 
> > Adapt the script to be one-based, as well, and adapt the GitHub workflow
> > accordingly. Note that GitLab doesn't yet use the script, so it does not
> > need to be adapted. This will change in the next commit though.
> 
> This seems to require the following to be squashed in:
> 
> -- snip --
> From 677370c8cd3b73358713a0efac6bb573b347d53f Mon Sep 17 00:00:00 2001
> From: Johannes Schindelin <johannes.schindelin@gmx.de>
> Date: Wed, 18 Feb 2026 22:12:27 +0100
> Subject: [PATCH] fixup??? ci: make test slicing consistent across Meson/Make
> 
> This seems to be needed to let the first "win+VS test" job pass. See
> https://github.com/git-for-windows/git/actions/runs/22153194896/job/64053953314#step:5:70
> ("fatal: slice must be in the range 1 <= slice <= 10, got '0'").

Huh, weird, why didn't I spot this myself? Oh, these tests don't execute
by default, right? That's why.

> Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
> ---
>  .github/workflows/main.yml | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/.github/workflows/main.yml b/.github/workflows/main.yml
> index 412842eb3e2..42585fff779 100644
> --- a/.github/workflows/main.yml
> +++ b/.github/workflows/main.yml
> @@ -240,7 +240,7 @@ jobs:
>        shell: bash
>        env:
>          NO_SVN_TESTS: 1
> -      run: . /etc/profile && ci/run-test-slice.sh ${{matrix.nr}} 10
> +      run: . /etc/profile && ci/run-test-slice.sh $((${{matrix.nr}}+1)) 10
>      - name: print test failures
>        if: failure() && env.FAILED_TEST_ARTIFACTS != ''
>        shell: bash
> -- snap --
> 
> This here run proves that this change fixes the CI failure:
> https://github.com/git-for-windows/git/actions/runs/22157880378/job/64067214717#step:5:70
> 
> Patrick, would you mind adopting this amendment into your patch?

Yup, looks obviously correct. Will squash in, thanks!

Patrick

[PATCH 6/7] gitlab-ci: use "run-test-slice-meson.sh"

[Patrick Steinhardt]

While our GitHub workflow already uses "ci/run-test-slice-meson.sh",
GitLab CI open-codes the parameters. Adapt the latter to also use the
same script so that we always use the same Meson options across both CI
systems.

Signed-off-by: Patrick Steinhardt <ps@pks.im>
---
 .gitlab-ci.yml | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
index b419a84e2c..04857b479d 100644
--- a/.gitlab-ci.yml
+++ b/.gitlab-ci.yml
@@ -183,7 +183,8 @@ test:msvc-meson:
     - job: "build:msvc-meson"
       artifacts: true
   script:
-    - meson test -C build --no-rebuild --print-errorlogs --slice $Env:CI_NODE_INDEX/$Env:CI_NODE_TOTAL
+    - |
+      & "C:/Program Files/Git/usr/bin/bash.exe" -l -c 'ci/run-test-slice-meson.sh build $CI_NODE_INDEX $CI_NODE_TOTAL'
   parallel: 10
   artifacts:
     reports:

-- 
2.53.0.295.g64333814d3.dirty

[PATCH 7/7] gitlab-ci: handle failed tests on MSVC+Meson job

[Patrick Steinhardt]

The MSVC+Meson job does not currently have any logic to print failing
tests, nor does it upload the failed test artifacts. Backfill this logic
to make help debugging efforts in case any of its jobs has failed.

GitHub already knows to do this, so we don't need an equivalent change
over there.

Signed-off-by: Patrick Steinhardt <ps@pks.im>
---
 .gitlab-ci.yml | 14 +++++++++++++-
 1 file changed, 13 insertions(+), 1 deletion(-)

diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
index 04857b479d..71b8a6e642 100644
--- a/.gitlab-ci.yml
+++ b/.gitlab-ci.yml
@@ -157,6 +157,8 @@ test:mingw64:
   parallel: 10
 
 .msvc-meson:
+  variables:
+    TEST_OUTPUT_DIRECTORY: "C:/Git-Test"
   tags:
     - saas-windows-medium-amd64
   before_script:
@@ -164,12 +166,13 @@ test:mingw64:
     - choco install -y git meson ninja rust-ms
     - Import-Module $env:ChocolateyInstall\helpers\chocolateyProfile.psm1
     - refreshenv
+    - New-Item -Path $env:TEST_OUTPUT_DIRECTORY -ItemType Directory
 
 build:msvc-meson:
   extends: .msvc-meson
   stage: build
   script:
-    - meson setup build --vsenv -Dperl=disabled -Dbackend_max_links=1 -Dcredential_helpers=wincred
+    - meson setup build --vsenv -Dperl=disabled -Dbackend_max_links=1 -Dcredential_helpers=wincred -Dtest_output_directory="$TEST_OUTPUT_DIRECTORY"
     - meson compile -C build
   artifacts:
     paths:
@@ -185,10 +188,19 @@ test:msvc-meson:
   script:
     - |
       & "C:/Program Files/Git/usr/bin/bash.exe" -l -c 'ci/run-test-slice-meson.sh build $CI_NODE_INDEX $CI_NODE_TOTAL'
+  after_script:
+    - |
+      if ($env:CI_JOB_STATUS -ne "success") {
+        & "C:/Program Files/Git/usr/bin/bash.exe" -l -c 'ci/print-test-failures.sh'
+        Move-Item -Path "$env:TEST_OUTPUT_DIRECTORY/failed-test-artifacts" -Destination t/
+      }
   parallel: 10
   artifacts:
+    paths:
+      - t/failed-test-artifacts
     reports:
       junit: build/meson-logs/testlog.junit.xml
+    when: on_failure
 
 test:fuzz-smoke-tests:
   image: ubuntu:latest

-- 
2.53.0.295.g64333814d3.dirty

Re: [PATCH 0/7] Some assorted fixes for GitLab CI

[Patrick Steinhardt]

On Wed, Feb 11, 2026 at 11:34:26AM +0100, Patrick Steinhardt wrote:
> Changes in v2:
>   - Fix syntax in GitHub workflow.
>   - Add another patch on top that fixes win+Meson tests in GitHub. Those
>     tests didn't execute at all, even before this patch series.
>   - This uncovers another bug in how we set up MERGE_TOOLS_DIR, so this
>     results in another commit.
>   - I've now tested also tested the changes on GitHub [3].
>   - Link to v1: https://lore.kernel.org/r/20260209-b4-pks-ci-msvc-iconv-fixes-v1-0-1e3167cd8828@pks.im

Oops, sorry, I've screwed up this version a bit. This is supposed to be
v2 of the series at [1].

Patrick

[1]: <20260209-b4-pks-ci-meson-improvements-v1-0-38444dec4874@pks.im>

Re: [PATCH 0/7] Some assorted fixes for GitLab CI

[Junio C Hamano]

Patrick Steinhardt <ps@pks.im> writes:

> On Wed, Feb 11, 2026 at 11:34:26AM +0100, Patrick Steinhardt wrote:
>> Changes in v2:
>>   - Fix syntax in GitHub workflow.
>>   - Add another patch on top that fixes win+Meson tests in GitHub. Those
>>     tests didn't execute at all, even before this patch series.
>>   - This uncovers another bug in how we set up MERGE_TOOLS_DIR, so this
>>     results in another commit.
>>   - I've now tested also tested the changes on GitHub [3].
>>   - Link to v1: https://lore.kernel.org/r/20260209-b4-pks-ci-msvc-iconv-fixes-v1-0-1e3167cd8828@pks.im
>
> Oops, sorry, I've screwed up this version a bit. This is supposed to be
> v2 of the series at [1].
>
> Patrick
>
> [1]: <20260209-b4-pks-ci-meson-improvements-v1-0-38444dec4874@pks.im>

Will replace.  Thanks!

[PATCH v2 0/7] Some assorted fixes for GitLab CI

[Patrick Steinhardt]

Hi,

I recently had the pleasure of debugging a couple of failing
MSVC+Windows jobs in GitLab CI, which hasn't been quite fun because we
didn't know to print error logs, and neither did we upload the failed
test artifacts. This patch series is the result of this frustration and
fixes a couple of smaller issues in the context of our CI:

  - I noticed that test slicing is slightly wrong because of a
    difference between zero- and one-based indices, which causes us to
    skip the first test on GitLab.

  - I deduplicated how we run Meson tests so that both GitLab and GitHub
    use the same "run-test-slice-meson.sh" script.

  - I add logic to handle failing tests via "print-test-failures.sh".

The result can be found at [1]. Note that tests are failing, but those
failures are fixed in a separate patch series via [2]. In any case, I
guess those test failures also serve as a good demonstration how the
failing tests show up now.

Changes in v3:
  - Correct a second invocation of "ci/run-test-slice.sh". I missed this
    invocation, and because this job doesn't run by default I didn't
    spot that it broke.
  - Link to v1: https://lore.kernel.org/r/20260211-b4-pks-ci-meson-improvements-v1-0-cb167cc80b86@pks.im

Changes in v2:
  - Fix syntax in GitHub workflow.
  - Add another patch on top that fixes win+Meson tests in GitHub. Those
    tests didn't execute at all, even before this patch series.
  - This uncovers another bug in how we set up MERGE_TOOLS_DIR, so this
    results in another commit.
  - I've now tested also tested the changes on GitHub [3].
  - Link to v1: https://lore.kernel.org/r/20260209-b4-pks-ci-msvc-iconv-fixes-v1-0-1e3167cd8828@pks.im

Thanks!

Patrick

[1]: https://gitlab.com/gitlab-org/git/-/merge_requests/497
[2]: <20260209-b4-pks-ci-msvc-iconv-fixes-v1-0-1e3167cd8828@pks.im>
[3]: https://github.com/git/git/pull/2195

---
Patrick Steinhardt (7):
      ci: handle failures of test-slice helper
      ci: don't skip smallest test slice in GitLab
      meson: fix MERGE_TOOL_DIR with "--no-bin-wrappers"
      github: fix Meson tests not executing at all
      ci: make test slicing consistent across Meson/Make
      gitlab-ci: use "run-test-slice-meson.sh"
      gitlab-ci: handle failed tests on MSVC+Meson job

 .github/workflows/main.yml |  8 ++++----
 .gitlab-ci.yml             | 17 +++++++++++++++--
 ci/run-test-slice-meson.sh |  2 +-
 ci/run-test-slice.sh       |  6 +++---
 t/helper/test-path-utils.c | 18 ++++++++++++------
 t/meson.build              |  1 +
 6 files changed, 36 insertions(+), 16 deletions(-)

Range-diff versus v1:

1:  b5ad91e1bc = 1:  b55e0b145f ci: handle failures of test-slice helper
2:  5bed073cfc ! 2:  fbdd1df05d ci: don't skip smallest test slice in GitLab
    @@ Commit message
         As the indices are now one-based it means that GitLab's CI is fixed.
         The GitHub workflow is updated accordingly.
     
    +    Helped-by: Johannes Schindelin <johannes.schindelin@gmx.de>
         Signed-off-by: Patrick Steinhardt <ps@pks.im>
     
      ## .github/workflows/main.yml ##
    @@ .github/workflows/main.yml: jobs:
          - name: test
            shell: bash
     -      run: . /etc/profile && ci/run-test-slice.sh ${{matrix.nr}} 10
    ++      run: . /etc/profile && ci/run-test-slice.sh $((${{matrix.nr}} + 1)) 10
    +     - name: print test failures
    +       if: failure() && env.FAILED_TEST_ARTIFACTS != ''
    +       shell: bash
    +@@ .github/workflows/main.yml: jobs:
    +       shell: bash
    +       env:
    +         NO_SVN_TESTS: 1
    +-      run: . /etc/profile && ci/run-test-slice.sh ${{matrix.nr}} 10
     +      run: . /etc/profile && ci/run-test-slice.sh $((${{matrix.nr}} + 1)) 10
          - name: print test failures
            if: failure() && env.FAILED_TEST_ARTIFACTS != ''
3:  6b061b829d = 3:  a5e411aade meson: fix MERGE_TOOL_DIR with "--no-bin-wrappers"
4:  828147d641 = 4:  b7d868fc53 github: fix Meson tests not executing at all
5:  05458adc6f = 5:  c9690b10a8 ci: make test slicing consistent across Meson/Make
6:  49df181d3d = 6:  bee3a9d944 gitlab-ci: use "run-test-slice-meson.sh"
7:  e7e9504c16 = 7:  ea1e1a4d87 gitlab-ci: handle failed tests on MSVC+Meson job

---
base-commit: 3e0db84c88c57e70ac8be8c196dfa92c5d656fbc
change-id: 20260209-b4-pks-ci-meson-improvements-93d8a1ffdd27

[PATCH v2 1/7] ci: handle failures of test-slice helper

[Patrick Steinhardt]

The "run-test-slice.sh" script executes the test helper to slice up
tests passed to it. As the execution is part of a pipe though, we end up
ignoring any potential error code returned by the helper.

Make the code more robust by storing the tests in a variable first so
that we can split up the pipeline.

Signed-off-by: Patrick Steinhardt <ps@pks.im>
---
 ci/run-test-slice.sh | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/ci/run-test-slice.sh b/ci/run-test-slice.sh
index 0444c79c02..ff948e397f 100755
--- a/ci/run-test-slice.sh
+++ b/ci/run-test-slice.sh
@@ -5,9 +5,9 @@
 
 . ${0%/*}/lib.sh
 
-group "Run tests" make --quiet -C t T="$(cd t &&
-	./helper/test-tool path-utils slice-tests "$1" "$2" t[0-9]*.sh |
-	tr '\n' ' ')" ||
+TESTS=$(cd t && ./helper/test-tool path-utils slice-tests "$1" "$2" t[0-9]*.sh)
+
+group "Run tests" make --quiet -C t T="$(echo "$TESTS" | tr '\n' ' ')" ||
 handle_failed_tests
 
 # We only have one unit test at the moment, so run it in the first slice

-- 
2.53.0.414.gf7e9f6c205.dirty

[PATCH v2 2/7] ci: don't skip smallest test slice in GitLab

[Patrick Steinhardt]

The "ci/run-test-slice.sh" script can be used to slice up all of our
tests into N pieces and then run each of them on a separate CI job.
This is used by both GitLab and GitHub CI to speed up Windows tests,
which would otherwise be painfully slow.

The infra itself is fueled by `test-tool path-utils slice-tests`. This
tool receives as input an "offset" and a "stride" that can be combined
to slice up tests. This framing can be misleading though: you are
expected to pass a zero-based index as "offset", and the complete number
of slices to the "stride". The latter makes sense, but it is somewhat
surprising that the offset needs to be zero-based. And this is in fact
biting us: while GitHub passes zero-based indices, GitLab passes
`$CI_NODE_INDEX`, which is a one-based indice.

Ideally, we should have verification that the parameters make sense.
And naturally, one would for example expect that it's an error to call
the binary with an offset larger than the stride. But with the current
framing as "offset" it's not even wrong to do so, as it is of course
well-defined to start at a larger offset than the stride.

This means that we get this wrong on GitLab's CI, as we pass a one based
index there, and this causes us to skip one of the tests. Interestingly,
it's not the lexicographically first test that we skip. Instead, as we
sort tests by size before slicing them, we skip the _smallest_ test.

Reframe the problem to instead talk about "slice number" and "total
number of slices". For all of our use cases this is semantically
equivalent, but it allows us to perform some verifications:

  - The total number of slices must be greater than 1.

  - The selected slice must be between 1 <= nr <= slices_total.

As the indices are now one-based it means that GitLab's CI is fixed.
The GitHub workflow is updated accordingly.

Helped-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Patrick Steinhardt <ps@pks.im>
---
 .github/workflows/main.yml |  4 ++--
 t/helper/test-path-utils.c | 18 ++++++++++++------
 2 files changed, 14 insertions(+), 8 deletions(-)

diff --git a/.github/workflows/main.yml b/.github/workflows/main.yml
index f2e93f5461..ec1a660981 100644
--- a/.github/workflows/main.yml
+++ b/.github/workflows/main.yml
@@ -150,7 +150,7 @@ jobs:
     - uses: git-for-windows/setup-git-for-windows-sdk@v1
     - name: test
       shell: bash
-      run: . /etc/profile && ci/run-test-slice.sh ${{matrix.nr}} 10
+      run: . /etc/profile && ci/run-test-slice.sh $((${{matrix.nr}} + 1)) 10
     - name: print test failures
       if: failure() && env.FAILED_TEST_ARTIFACTS != ''
       shell: bash
@@ -237,7 +237,7 @@ jobs:
       shell: bash
       env:
         NO_SVN_TESTS: 1
-      run: . /etc/profile && ci/run-test-slice.sh ${{matrix.nr}} 10
+      run: . /etc/profile && ci/run-test-slice.sh $((${{matrix.nr}} + 1)) 10
     - name: print test failures
       if: failure() && env.FAILED_TEST_ARTIFACTS != ''
       shell: bash
diff --git a/t/helper/test-path-utils.c b/t/helper/test-path-utils.c
index f5f33751da..874542ec34 100644
--- a/t/helper/test-path-utils.c
+++ b/t/helper/test-path-utils.c
@@ -477,14 +477,20 @@ int cmd__path_utils(int argc, const char **argv)
 
 	if (argc > 5 && !strcmp(argv[1], "slice-tests")) {
 		int res = 0;
-		long offset, stride, i;
+		long slice, slices_total, i;
 		struct string_list list = STRING_LIST_INIT_NODUP;
 		struct stat st;
 
-		offset = strtol(argv[2], NULL, 10);
-		stride = strtol(argv[3], NULL, 10);
-		if (stride < 1)
-			stride = 1;
+		slices_total = strtol(argv[3], NULL, 10);
+		if (slices_total < 1)
+			die("there must be at least one slice, got '%s'",
+			    argv[3]);
+
+		slice = strtol(argv[2], NULL, 10);
+		if (1 > slice || slice > slices_total)
+			die("slice must be in the range 1 <= slice <= %ld, got '%s'",
+			    slices_total, argv[2]);
+
 		for (i = 4; i < argc; i++)
 			if (stat(argv[i], &st))
 				res = error_errno("Cannot stat '%s'", argv[i]);
@@ -492,7 +498,7 @@ int cmd__path_utils(int argc, const char **argv)
 				string_list_append(&list, argv[i])->util =
 					(void *)(intptr_t)st.st_size;
 		QSORT(list.items, list.nr, cmp_by_st_size);
-		for (i = offset; i < list.nr; i+= stride)
+		for (i = slice - 1; i < list.nr; i+= slices_total)
 			printf("%s\n", list.items[i].string);
 
 		return !!res;

-- 
2.53.0.414.gf7e9f6c205.dirty

[PATCH v2 3/7] meson: fix MERGE_TOOL_DIR with "--no-bin-wrappers"

[Patrick Steinhardt]

On Windows, we execute tests with "--no-bin-wrappers". This has been
introduced via a87e427e35 (ci: speed up Windows phase, 2019-01-29) to
save some time: spawning processes is expensive on Windows, and shell
scripts tend to spawn a bunch of them. So overall, the bin-wrappers led
to a performance overhead of ~10-30%.

This causes test failures when using Meson on Windows:

  failure: t7610.28 mergetool --tool-help shows recognized tools
    ++ git mergetool --tool-help
    /d/a/git/git/build/git-mergetool--lib: line 45: cd: D:/a/git/git/build/mergetools: No such file or directory

The root cause here is that our bin-wrappers are usually responsible for
setting up the `MERGE_TOOL_DIR` environment variable so that we can
locate these scripts. But as we don't use the bin-wrappers, we'll
instead use the default location for merge tools, which is derived from
`GIT_EXEC_PATH`. And as `GIT_EXEC_PATH` points to our build directory,
which won't ever contain any of the merge tools, we will fail to locate
any of the merge tools.

This issue has went unnoticed for a long time given that we only skip
bin-wrappers on Windows, and because the CI jobs on Windows didn't
execute due to a bug.

Fix the issue by always setting the `MERGE_TOOL_DIR` environment
variable to the correct directory.

Signed-off-by: Patrick Steinhardt <ps@pks.im>
---
 t/meson.build | 1 +
 1 file changed, 1 insertion(+)

diff --git a/t/meson.build b/t/meson.build
index 19e8306298..5ba7efd603 100644
--- a/t/meson.build
+++ b/t/meson.build
@@ -1208,6 +1208,7 @@ endif
 
 test_environment = script_environment
 test_environment.set('GIT_BUILD_DIR', git_build_dir)
+test_environment.set('MERGE_TOOLS_DIR', meson.project_source_root() / 'mergetools')
 
 foreach integration_test : integration_tests
   test(fs.stem(integration_test), shell,

-- 
2.53.0.414.gf7e9f6c205.dirty

[PATCH v2 4/7] github: fix Meson tests not executing at all

[Patrick Steinhardt]

While the win+Meson test jobs run in GitHub workflows, the shell script
that is supposed to run the jobs is seemingly not running at all. All
that the CI job prints is the following:

  Run ci/run-test-slice-meson.sh build 1 10
    ci/run-test-slice-meson.sh build 1 10
    shell: C:\Program Files\PowerShell\7\pwsh.EXE -command ". '{0}'"
    env:
      DEVELOPER: 1

The step is currently defined to use PowerShell, and of course it
doesn't know how to execute POSIX shell scripts. What's surprising
though is that this step doesn't even lead to a CI failure.

Fix the issue by using Bash instead of PowerShell, as we do in other
steps that execute shell scripts.

Signed-off-by: Patrick Steinhardt <ps@pks.im>
---
 .github/workflows/main.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/workflows/main.yml b/.github/workflows/main.yml
index ec1a660981..a011d8d0f9 100644
--- a/.github/workflows/main.yml
+++ b/.github/workflows/main.yml
@@ -297,7 +297,7 @@ jobs:
         name: windows-meson-artifacts
         path: build
     - name: Test
-      shell: pwsh
+      shell: bash
       run: ci/run-test-slice-meson.sh build ${{matrix.nr}} 10
     - name: print test failures
       if: failure() && env.FAILED_TEST_ARTIFACTS != ''

-- 
2.53.0.414.gf7e9f6c205.dirty

[PATCH v2 5/7] ci: make test slicing consistent across Meson/Make

[Patrick Steinhardt]

In the preceding commit we have adjusted test slicing to be one-based
when using the "ci/run-test-slice.sh" script. But we also have an
equivalent script for Meson that is still zero-based, which is of course
inconsistent.

Adapt the script to be one-based, as well, and adapt the GitHub workflow
accordingly. Note that GitLab doesn't yet use the script, so it does not
need to be adapted. This will change in the next commit though.

Signed-off-by: Patrick Steinhardt <ps@pks.im>
---
 .github/workflows/main.yml | 2 +-
 ci/run-test-slice-meson.sh | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/.github/workflows/main.yml b/.github/workflows/main.yml
index a011d8d0f9..826f2f5d3a 100644
--- a/.github/workflows/main.yml
+++ b/.github/workflows/main.yml
@@ -298,7 +298,7 @@ jobs:
         path: build
     - name: Test
       shell: bash
-      run: ci/run-test-slice-meson.sh build ${{matrix.nr}} 10
+      run: ci/run-test-slice-meson.sh build $((${{matrix.nr}} + 1)) 10
     - name: print test failures
       if: failure() && env.FAILED_TEST_ARTIFACTS != ''
       shell: bash
diff --git a/ci/run-test-slice-meson.sh b/ci/run-test-slice-meson.sh
index 961c94fba0..a6df927ba5 100755
--- a/ci/run-test-slice-meson.sh
+++ b/ci/run-test-slice-meson.sh
@@ -9,5 +9,5 @@
 
 group "Run tests" \
 	meson test -C "$1" --no-rebuild --print-errorlogs \
-		--test-args="$GIT_TEST_OPTS" --slice "$((1+$2))/$3" ||
+		--test-args="$GIT_TEST_OPTS" --slice "$(($2))/$3" ||
 handle_failed_tests

-- 
2.53.0.414.gf7e9f6c205.dirty

[PATCH v2 6/7] gitlab-ci: use "run-test-slice-meson.sh"

[Patrick Steinhardt]

While our GitHub workflow already uses "ci/run-test-slice-meson.sh",
GitLab CI open-codes the parameters. Adapt the latter to also use the
same script so that we always use the same Meson options across both CI
systems.

Signed-off-by: Patrick Steinhardt <ps@pks.im>
---
 .gitlab-ci.yml | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
index b419a84e2c..04857b479d 100644
--- a/.gitlab-ci.yml
+++ b/.gitlab-ci.yml
@@ -183,7 +183,8 @@ test:msvc-meson:
     - job: "build:msvc-meson"
       artifacts: true
   script:
-    - meson test -C build --no-rebuild --print-errorlogs --slice $Env:CI_NODE_INDEX/$Env:CI_NODE_TOTAL
+    - |
+      & "C:/Program Files/Git/usr/bin/bash.exe" -l -c 'ci/run-test-slice-meson.sh build $CI_NODE_INDEX $CI_NODE_TOTAL'
   parallel: 10
   artifacts:
     reports:

-- 
2.53.0.414.gf7e9f6c205.dirty

[PATCH v2 7/7] gitlab-ci: handle failed tests on MSVC+Meson job

[Patrick Steinhardt]

The MSVC+Meson job does not currently have any logic to print failing
tests, nor does it upload the failed test artifacts. Backfill this logic
to make help debugging efforts in case any of its jobs has failed.

GitHub already knows to do this, so we don't need an equivalent change
over there.

Signed-off-by: Patrick Steinhardt <ps@pks.im>
---
 .gitlab-ci.yml | 14 +++++++++++++-
 1 file changed, 13 insertions(+), 1 deletion(-)

diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
index 04857b479d..71b8a6e642 100644
--- a/.gitlab-ci.yml
+++ b/.gitlab-ci.yml
@@ -157,6 +157,8 @@ test:mingw64:
   parallel: 10
 
 .msvc-meson:
+  variables:
+    TEST_OUTPUT_DIRECTORY: "C:/Git-Test"
   tags:
     - saas-windows-medium-amd64
   before_script:
@@ -164,12 +166,13 @@ test:mingw64:
     - choco install -y git meson ninja rust-ms
     - Import-Module $env:ChocolateyInstall\helpers\chocolateyProfile.psm1
     - refreshenv
+    - New-Item -Path $env:TEST_OUTPUT_DIRECTORY -ItemType Directory
 
 build:msvc-meson:
   extends: .msvc-meson
   stage: build
   script:
-    - meson setup build --vsenv -Dperl=disabled -Dbackend_max_links=1 -Dcredential_helpers=wincred
+    - meson setup build --vsenv -Dperl=disabled -Dbackend_max_links=1 -Dcredential_helpers=wincred -Dtest_output_directory="$TEST_OUTPUT_DIRECTORY"
     - meson compile -C build
   artifacts:
     paths:
@@ -185,10 +188,19 @@ test:msvc-meson:
   script:
     - |
       & "C:/Program Files/Git/usr/bin/bash.exe" -l -c 'ci/run-test-slice-meson.sh build $CI_NODE_INDEX $CI_NODE_TOTAL'
+  after_script:
+    - |
+      if ($env:CI_JOB_STATUS -ne "success") {
+        & "C:/Program Files/Git/usr/bin/bash.exe" -l -c 'ci/print-test-failures.sh'
+        Move-Item -Path "$env:TEST_OUTPUT_DIRECTORY/failed-test-artifacts" -Destination t/
+      }
   parallel: 10
   artifacts:
+    paths:
+      - t/failed-test-artifacts
     reports:
       junit: build/meson-logs/testlog.junit.xml
+    when: on_failure
 
 test:fuzz-smoke-tests:
   image: ubuntu:latest

-- 
2.53.0.414.gf7e9f6c205.dirty

Re: [PATCH v2 0/7] Some assorted fixes for GitLab CI

[Junio C Hamano]

Patrick Steinhardt <ps@pks.im> writes:

> Changes in v3:
>   - Correct a second invocation of "ci/run-test-slice.sh". I missed this
>     invocation, and because this job doesn't run by default I didn't
>     spot that it broke.
>   - Link to v1: https://lore.kernel.org/r/20260211-b4-pks-ci-meson-improvements-v1-0-cb167cc80b86@pks.im

Thanks, all.  Let's mark the topic for 'next'.