On Thursday, August 1, 2024 10:44 AM Patrick Steinhardt wrote: >On Thu, Aug 01, 2024 at 10:04:37AM -0400, rsbecker@xxxxxxxxxxxxx wrote: >> On Thursday, August 1, 2024 9:55 AM, Patrick Steinhardt wrote: >> >On Thu, Aug 01, 2024 at 09:53:52AM -0400, rsbecker@xxxxxxxxxxxxx wrote: >> >> On Thursday, August 1, 2024 9:50 AM, Patrick Steinhardt wrote: >> >> >On Thu, Aug 01, 2024 at 09:47:38AM -0400, rsbecker@xxxxxxxxxxxxx >wrote: >> >> >> On Thursday, August 1, 2024 9:37 AM, Patrick Steinhardt wrote: >> >> >> >As mentioned in another mail, we do not use its Makefile at all. >> >> >> >Did you >> >> >> check >> >> >> >whether the version I have proposed here works when running >> >> >> >`make >> test`? >> >> >> >> >> >> That is the commit I have been trying to use. make test in clar >> >> >> or >> git? >> >> > >> >> >In Git itself. `make test` builds and runs our unit tests, and >> >> >that now >> >> also includes >> >> >unit tests based on clar with this patch series. The clar Makefile >> >> >exists >> >> only because I >> >> >did a 1:1 import of the upstream dependency. We could just as well >> >> >remove >> >> it >> >> >altogether, including other bits that we don't end up using. >> >> >> >> I see. Well, the 2.46.0 test passes. I ran 'seen' 5 days ago and >> >> 'next' is going now. Would that catch it? >> > >> >No, as the patches in this thread are only up for discussion right >> >now and >> have not >> >been merged to any of the branches. You'd have to apply them on top >> >of >> v2.46.0 >> >first :) >> >> Do you happen to have a public fork? > >You can pull the branch pks-clar-unit-tests from https://gitlab.com/gitlab-org/git. Fix #1: diff --git a/t/unit-tests/clar/clar/sandbox.h b/t/unit-tests/clar/clar/sandbox.h index 0ba1479620..81a40f26b8 100644 --- a/t/unit-tests/clar/clar/sandbox.h +++ b/t/unit-tests/clar/clar/sandbox.h @@ -116,7 +116,7 @@ static int build_sandbox_path(void) strncpy(_clar_path + len, path_tail, sizeof(_clar_path) - len); -#if defined(__MINGW32__) +#if defined(__MINGW32__) || defined(__TANDEM) if (_mktemp(_clar_path) == NULL) return -1; That gets past the clar build problem. Probably more to come as I encounter issues. The assert issue was just a warning about a constant expression that will always fail.
