On Mon, Jul 12, 2010 at 5:48 PM, Joel Rosdahl <j...@rosdahl.net> wrote: > Regarding the "Win32 support" patch: > > I suspect that the static inline functions in ccache.h will result in > portibility problems, so I think it's better to just move them to util.c.
The semester at university is over, so now I have time to look at the patches again in the next few days. > Maybe the definition and use of is_win32_dir should be surrounded by > #ifdef _WIN32 guards? I've reworked the mmap wrapper to avoid the need of is_win32_dir (it caused a failure in cygwin the way it was). I'll update my git fork and ping the list when it's done. > Why is a check for both "ccache" and "ccache.exe" needed? Doesn't > argv[0] on Windows always end with the actual extension? The reason I'm > asking is that if only "ccache.exe" is needed, we could maybe use the > EXEEXT variable from autoconf and avoid the conditional compilation. > > Why is the special handling of shell scripts needed? It had something to do with the regression tests. When the compiler being called is a script around the actual compiler, CreateProcess fails. [...] Do you think you could update the list configuration so that Reply-To points to the list instead of the sender? Thanks, Ramiro Polla _______________________________________________ ccache mailing list ccache@lists.samba.org https://lists.samba.org/mailman/listinfo/ccache
