On 2019-04-08 13:16, Sven Strickroth wrote: > A path such as 'c:/somepath/submodule/../.git/modules/submodule' wasn't > resolved correctly any more, because the *nix variant of > offset_1st_component is used instead of the Win32 specific version. > > Regression was introduced in commit > 25d90d1cb72ce51407324259516843406142fe89. Was it ? 25d90d1cb merged this commit: 1cadad6f6 (junio/tb/use-common-win32-pathfuncs-on-cygwin) And, if I read that correctly, 1cadad6f6 does not change anything for MSVC. And the problem with the missing/wrong path resolution was there before 1cadad6f6 and after 1cadad6f6. >From that point of view, the patch looks correct, but: The other question: In config.mak.uname we need to add a line compat/win32/path-utils.o for the Windows build. In the git-for windows codebase I see COMPAT_OBJS +=compat/win32/path-utils 3 times: For Cygwin, MINGW and Windows. In git.git only for Cygwin and MINGW. (I don't have MSVC, so I can't test) > > Signed-off-by: Sven Strickroth <email@xxxxxxxxxx> > --- > git-compat-util.h | 1 + > 1 file changed, 1 insertion(+) > > diff --git a/git-compat-util.h b/git-compat-util.h > index e0275da7e0..9be177e588 100644 > --- a/git-compat-util.h > +++ b/git-compat-util.h > @@ -210,6 +210,7 @@ > #include "compat/mingw.h" > #include "compat/win32/fscache.h" > #elif defined(_MSC_VER) > +#include "compat/win32/path-utils.h" > #include "compat/msvc.h" > #include "compat/win32/fscache.h" > #else >
