This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [Patch] [mingw32] fix regresion with trailing backslashes in -Iinclude dir


Danny Smith <dannysmith@clear.net.nz> writes:

> From: "Zack Weinberg" <zack@codesourcery.com>
>
>
> | Danny Smith <dannysmith@clear.net.nz> writes:
> | 
> | > Hello,
> | >
> | > On mingw32, with a command-line -I include option ending with a
> | > backslash gcc fails to recognise an existing directory. For example,
> | > '-I..\include\' reports a non-existent directory while '-I..\include' works.
> | > This is a regression from 3.3.3 and is observed on both 3.4 and and
> | > trunk.
> | >
> | > Trailing forward slashes cause no problems.
> | >
> | > For some strange reason, the stat function in MS runtime accepts
> | > POSIX-style '/' but not DOS-style '\\' as final character for a
> | > directory name 
> | ...
> | 
> | >         * c-incpath.c (add_path): Canonicalize paths to use '/' if
> | >         HAVE_DOS_BASED_FILESYSTEM.
> | 
> | This is OK, but please add a comment explaining what you just
> | discovered, so someone doesn't take it back out again.
> | 
>
> Is this sufficient?  Is OK for trunk only or for 3.4.1 as well.  

Yes, that's sufficient.  Being a regression from 3.3 it is OK for
3.4.1.

zw


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]