From: Ruben V. B. <van...@gm...> - 2010-06-15 13:49:30
|
2010/6/15 JonY <jo...@us...> > On 6/15/2010 20:30, Ruben Van Boxem wrote: > >> 2010/6/15 JonY<jo...@us...> >> >> On 6/15/2010 19:31, Ruben Van Boxem wrote: >>> >>> I didn't have such issues cross compiling GMP in Cygwin or Linux. >>>> >>>>> >>>>> >>>> >>>> Well, MSYS is special :s (see other thread on my link.exe removal). >>>> >>>> >>>> MSYS is not special at all to configure scripts. How does your PATH >>> look? I >>> am very puzzled how this can happen. >>> >>> >> I know this, I meant the precautions and gotcha's MSYS sometimes has, like >> the link.exe issue we are having here. >> My PATH from within MSYS is this: >> .:/usr/local/bin:/mingw/bin:/bin:/c/Program Files (x86)/MiKTeX >> >> 2.8/miktex/bin:/c/Windows/system32:/c/Windows:/c/Windows/System32/Wbem:/c/Windows/System32/WindowsPowerShell/v1.0:/c/Program >> Files/MATLAB/R2010a/runtime/win64:/c/Program >> Files/MATLAB/R2010a/bin:/c/Program Files (x86)/MKVtoolnix >> >> I'm guessing /usr/local/bin is /bin, and mingw/bin is set by the >> postinstall >> script to wherever you have a mingw compiler installed >> >> > Hmm, I think /mingw/bin or where your mingw-w64 toolchain bindir is should > always be before MSYS's bindir, and that "." will certainly cause trouble > with GCC's build system. > That's exactly why I said MSYS is special... and I don't know what will break if I try to fix it. I still find it very odd that GMP configure even knows about the name "link", because no mingw toolchain even has such a thing. Shouldn't the script use ld or something else for its tests? |