similar to: [LLVMdev] GCC_NO_EXECUTABLES during CFE build on MinGW

Hi, I would like this patch for Path.inc to be accepted. I cannot tell, why the win32 version of isValid() method fails when issuing a correct path on the mingw platform. This is especially annoying when configuring the cfrontend, because we get this misindicating error: 'checking for sin in -lm... configure: error: Link tests are not allowed after GCC_NO_EXECUTABLES.' And it has

Hi, John Chriswell wrote: >You might want to google for AC_NO_EXECUTABLES. It's an autoconf test that >generates the >behavior you describe. Apparently the GCC folks wrote a new >version of the macro due to the >above behavior. As suggested, I googled and found this interesting answer for above error and behavior: http://www.mail-archive.com/gcc-bugs at

Hello, Greg. You wrote Wednesday, May 10, 2006, 3:40:43 PM: GP> checking for main in -lm... configure: error: Link tests are not allowed GP> after GCC_NO_EXECUTABLES. If my memory serves me right, this message was due to something odd happened in gccld. Maybe supplied path was wrong or something like this. I've seen this bug before, but don't remember how I've fixed it. --

I've blown a couple of weekends trying to build under MinGW. I can build the llvm tools with only a couple of minor hassles, but can't seem to make any headway with the C Front End Most errors I run into involve not being able to find header files. The most frustrating part is that the error messages report that the file can't be found, and then proceed to give a path to where the

I'm using a little shell script: BUILD_ROOT=/home/llvm-1.7/cfrontend PREFIX="$BUILD_ROOT/install" LOCAL_BUILD_DIR="$BUILD_ROOT/build" SOURCE_DIR="$BUILD_ROOT/src" echo $__me: Building $TARGET echo $__me: BUILD_ROOT == $BUILD_ROOT echo $__me: SOURCE_DIR == $SOURCE_DIR echo $__me: LOCAL_BUILD_DIR == $LOCAL_BUILD_DIR echo $__me: PREFIX == $PREFIX

On Sat, 20 Nov 2004, Henrik Bach wrote: > I can't figure what this means: > ------------------ > ... > checking for g++ that supports -ffunction-sections -fdata-sections... no > checking for sin in -lm... configure: error: Link tests are not allowed > after GCC_NO_EXECUTABLES. > make: *** [configure-target-libstdc++-v3] Error 1 > ------------------ I've never seen

Hi, I can't figure what this means: ------------------ ... checking for g++ that supports -ffunction-sections -fdata-sections... no checking for sin in -lm... configure: error: Link tests are not allowed after GCC_NO_EXECUTABLES. make: *** [configure-target-libstdc++-v3] Error 1 ------------------ Henrik. ============================================================= Henrik Bach Open Source

Is the MinGW user's list on the web? Can you give me a link to the thread? (What's IIRC?) Thanks. On Sat, 22 Apr 2006, Oscar Fuentes wrote: > Greg Pettyjohn <gregp at ccs.neu.edu> writes: > > > I've blown a couple of weekends trying to build under MinGW. > > > > I can build the llvm tools with only a couple of minor hassles, but can't > > seem

Greg Pettyjohn <gregp at ccs.neu.edu> writes: > Is the MinGW user's list on the web? Look on www.mingw.org -> mailing list > Can you give me a link to the thread? Sorry, no time for that. Once on the mailing list archives, do a search for 'configure'. > (What's IIRC?) If I Remember Correctly. By the way, what's your configure command? -- Oscar

Greg Pettyjohn <gregp at ccs.neu.edu> writes: > I've blown a couple of weekends trying to build under MinGW. > > I can build the llvm tools with only a couple of minor hassles, but can't > seem to make any headway with the C Front End > > Most errors I run into involve not being able to find header files. > The most frustrating part is that the error messages

Thanks. I'll try some of this. Further comments in-line: On Sun, 23 Apr 2006, Oscar Fuentes wrote: > Greg Pettyjohn <gregp at ccs.neu.edu> writes: > > > I'm using a little shell script: > [snip] > > confopts="--prefix=$PREFIX" > > confopts="$confopts --disable-threads" > > confopts="$confopts --disable-nls" >

Greg Pettyjohn <gregp at ccs.neu.edu> writes: > I'm using a little shell script: [snip] > confopts="--prefix=$PREFIX" > confopts="$confopts --disable-threads" > confopts="$confopts --disable-nls" > confopts="$confopts --disable-shared" > confopts="$confopts --enable-languages=c,c++" >

Thank you, I did try that and got the following error: checking for library containing strerror... configure: error: Link tests are not allowed after GCC_NO_EXECUTABLES. make[1]: *** [configure-target-libiberty] Error 1 make[1]: Leaving directory `/home/ghoang/llvm-test/llvm-gcc-2.8-build' make: *** [all] Error 2 I also tried adding --disable-libiberty but make would fail even earlier

The problem with building the frontend on Windows is that gcc cannot be bootstrapped using Window's native compiler -- i.e. VC++ -- unlike every other platform. It can be built on Windows using gcc, of course, but even then only if the entire GNU environment is present. The real problem is that the bootstrapped frontend binaries built using gcc may not properly link with LLVM binaries built

So the fundamental problem I'm trying to solve is how get my cfrontend build configured properly. The usual symptom of an incorrect configuration is the inability to find various header files. Here's some example spew: $ /home/llvm_home/llvm-gcc-build/gcc/xgcc -v -B/home/llvm_home/llvm-gcc-build/gcc/ -Bc:/msys/1.0/mingw32/bin/ -Bc:/msys/1.0/mingw32/lib/ -isystem

On Tue, 2 Nov 2004, Jeff Cohen wrote: > The problem with building the frontend on Windows is that gcc cannot be > bootstrapped using Window's native compiler -- i.e. VC++ -- unlike every > other platform. It can be built on Windows using gcc, of course, but > even then only if the entire GNU environment is present. Yeah, annoying. Unfortunately we're not up to fixing GCC :)

Greg, Looks like Jeff Cohen already fixed this in CVS. Please update and give it another shot. Thanks, Reid. On Sat, 2006-05-06 at 16:43 -0400, Greg Pettyjohn wrote: > > On Sat, 6 May 2006, Greg Pettyjohn wrote: > > > Now I'm having problems with this: > > > > llvm-ar rc ./libgcc.a libgcc/./_muldi3.o <and-lots-more-.o-files...> > >

On Sat, 6 May 2006, Greg Pettyjohn wrote: > Now I'm having problems with this: > > llvm-ar rc ./libgcc.a libgcc/./_muldi3.o <and-lots-more-.o-files...> > C:\msys\1.0\home\llvm_home\install\bin\llvm-ar.exe: <invalid>: path is not valid > I've tracked this down in the debugger. It is indeed a bug. The problem is that Path::isValid() will reject a string

Chris Lattner wrote: > On Sat, 6 May 2006, Reid Spencer wrote: >> Looks like Jeff Cohen already fixed this in CVS. Please update and give >> it another shot. > > Is this fixing the issue, or just papering over it? Why are we trying > to load an archive member without a path set? We're not. It's just a sentry value. > > -Chris > >> On Sat,

On Sat, 6 May 2006, Reid Spencer wrote: > Looks like Jeff Cohen already fixed this in CVS. Please update and give > it another shot. Is this fixing the issue, or just papering over it? Why are we trying to load an archive member without a path set? -Chris > On Sat, 2006-05-06 at 16:43 -0400, Greg Pettyjohn wrote: >> >> On Sat, 6 May 2006, Greg Pettyjohn wrote: >>