Fix Windows-related -c options in match_baseline_files#1871
Fix Windows-related -c options in match_baseline_files#1871Byron merged 1 commit intoGitoxideLabs:mainfrom
-c options in match_baseline_files#1871Conversation
3fc0264 to
f8b3bb9
Compare
The `match_baseline_files.sh` fixture script adds entries to the index using `git update-index --add --cacheinfo`. These entries don't exist on disk, so it should be okay that some would cause problems on Windows if an attempt were made to actually create them. The tests that use the generated repository work on Windows. It is intended that the fixture script also be able to run on Windows, and Windows-related `-c` options are passed in commands where they are expected to be needed: `core.ignoreCase=false` and `core.protectNTFS=false`. But the fixture script nonetheless was failing on Windows because two commands that don't need `core.ignoreCase=false` but *do* need `core.protectNTFS=false` had `core.ignorecase=false` instead. This corrects that. This fixes two of the long-standing `GIX_TEST_IGNORE_ARCHIVES=1` Windows failures reported in GitoxideLabs#1358: - gix-pathspec::pathspec search::files - gix-pathspec::pathspec search::prefixes_are_always_case_sensitive
f8b3bb9 to
68bdccc
Compare
|
Thanks so much, this is great! I hope the auto-merge goes through now that the commit was rebased. |
The merge went through and it looks like everything is fine. But I'm not sure what got in the way before the last rebase. (From your wording, I presume there was a conflict or that some test failed.) To the best of my knowledge, the only other recent changes to a file changed here was in cdee7ff (added to #1882) and d340263 (in #1870) which reverted it. The changed file was So I don't know why there would have been a conflict. (I wonder if I made a mistake in the preceding rebase I did, from 3fc0264 to f8b3bb9, and used the wrong base, or something?) Anyway, I'm not worried about it. Please feel free to ignore this if you like. 😄 |
|
I don't think there was any conflicts for me, the rebase went through cleanly. Maybe that's the problem - a bad cherry-pick with unexpected results? |
I meant my own rebase that preceded yours. I did not suspect that your rebase (or manual cherry-pick) had any problems. I did check the diff that was merged here, though, and it looked fine. So as far as I can tell everything is okay. |
The
match_baseline_files.shfixture script adds entries to the index usinggit update-index --add --cacheinfo. These entries don't exist on disk, so it should be okay that some would cause problems on Windows if an attempt were made to actually create them. The tests that use the generated repository work on Windows.It is intended that the fixture script also be able to run on Windows, and Windows-related
-coptions are passed in commands where they are expected to be needed:core.ignoreCase=falseandcore.protectNTFS=false.But the fixture script nonetheless was failing on Windows because two commands that don't need
core.ignoreCase=falsebut do needcore.protectNTFS=falsehadcore.ignorecase=falseinstead. This corrects that.This fixes two of the long-standing
GIX_TEST_IGNORE_ARCHIVES=1Windows failures reported in #1358:gix-pathspec::pathspec search::filesgix-pathspec::pathspec search::prefixes_are_always_case_sensitiveThis includes a regenerated archive (generated on Arch Linux with Git 2.48.1) and updates
etc/test-fixtures-windows-expected-failures-see-issue-1358.txtto remove the two tests that now pass on Windows even withGIX_TEST_IGNORE_ARCHIVES=1.The
test-fixtures-windowsjob currently fails here due to #1849, the same as on main if CI is rerun there, as described in #1849 (comment). If #1870 is merged, the failures should go away. This PR could be rebased after merging if #1870, if desired.