The builds of gcc 4.3.3 do not produce a g77 or gfortran binary when emerged with the fortran useflag. Instead the binary gfortran-4.3.3 is produced. This breaks every package in the tree that uses fortran as none of the makefiles look for the fortran compiler in this location expecting it to be named either g77 or gfortran. Please fix the ebuild so the compiler is named sensibly. Reproducible: Always Steps to Reproduce: 1. install a gcc 4.3.3 with the fortran useflag (or without...) 2. try to install a package like cfitsio Actual Results: No compiler is found, a search of /usr/bin shows that there's no g77 or gfortran present. Expected Results: A g77 or gfortran exists and fortran packages in the tree are able to compile. I went crazy and marked this one as a blocker under the assumption that breaking every package in the tree written in fortran is probably a reasonably serious issue. Feel free to downgrade the severity if need be.
To clarify, there is a gfortran binary that is built and installed, it's just named gfortran-4.3.3 instead of gfortran. Packages expecting g77 will probably break even if this is properly named gfortran. (That said, I've symlinked this binary to g77 on my machine and it's working well enough to compile the fortran programs I've encountered in the tree so far...)
Well, from: 1. equery f =sys-devel/gcc-4.1.2: /usr/bin /usr/bin/c++-4.1.2 /usr/bin/cpp-4.1.2 /usr/bin/g++-4.1.2 /usr/bin/gcc-4.1.2 /usr/bin/gfortran-4.1.2 2. equery f =sys-devel/gcc-4.3.3: /usr/bin /usr/bin/c++-4.3.3 /usr/bin/cpp-4.3.3 /usr/bin/g++-4.3.3 /usr/bin/gcc-4.3.3 /usr/bin/gfortran-4.3.3 So this bug is probably invalid. Perhaps, something gone wrong during gcc-config ?
Yes, this is the case. A good round of gcc-config appears to have fixed this. Oops. Apologies, closing as invalid.