On 8/27/07, Stefan Esser <se_at_freebsd.org> wrote: > Maxim Khitrov wrote: > > > You got my hopes up ;-) No, I don't have that flag, but I had my > > CPUTYPE set to "native." Thought that was the problem, but > > unfortunately that didn't fix it. I first set it back to pentium-m, > > and then commented both CPUTYPE and CFLAGS in my make.conf. Same exact > > thing. Cannot compute suffix of object files. > > > > Any other ideas? > > > I had such a situation once before, and I'm quite sure, that it is > the compiler failing due to bad command line options. As a result > no output file is generated, configured can not determine the object > file name suffix and it terminates after printing the error message > you got. > > You can add an "echo $ac_compile" to just before the test that fails, > that way you should be able to see the command that is to be executed. > Then invoke the compiler (gcc-ooo !!!) just that same way (with any > C source file as additional parameter) and check the error message. > The error messages could in fact have been written to config.log, > but it was not in my failure case. > > Good luck, STefan Did as you suggested, put "echo $ac_compile" in configure, that printed out the compile command which was using $CFLAGS. I then had it run "echo $CFLAGS" and the arguments there contained -march=native. I guess this value was cached somewhere, and it was using it even though make.conf was changed. I didn't run make clean the first time since it takes a while to clean and rebuild everything up to that point. This time around I verified the settings in make.conf, then ran 'make clean install clean' for openoffice. Looks like it was able to build gcc-ooo with no problems. Right now it's building the actual openoffice installation. Thanks for the tip!Received on Mon Aug 27 2007 - 17:53:41 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:16 UTC