molden 4.8 is out. The proposed ebuild (see attachment) does not rely on fortran.eclass, adds some suggested optimizations, and turns on the mpfit module. Reproducible: Always Steps to Reproduce: This ebuild has been tested only on x86_64 architecture. It should work on other achitectures too, but it should be tested.
Created attachment 204336 [details] proposed ebuild for sci-chemistry/molden-4.8
Created attachment 204579 [details] modified ebuild -- fixes a buffer overflow condition After some testing, I've discovered a problem with any '-O?' option. This problem happens only when saving a file from Z matrix editor, determining a buffer overflow. To reproduce: 1) start molden 2) open the Z matrix editor 3) create some content 4) write Z matrix (to a file) -> buffer overflow detected. The issue is not present if no -O? options are present. Curiously, the bug is present in current portage ebuild too. this new ebuild fixes the issue. also, since -malign-double is on by default on x86_64 arch, the redundancy has been eliminated.
Thanks very much for your work and I've just added molden-4.8 to portage. The actual program abort is triggered by gcc's FORTIFY_SOURCE and caused by a fairly questionable strcpy in xwin.c which I've fixed with a patch. Since there are other similarly "suspicious" strcpy instances I'd not be too surprised if more of these FORTIFY_SOURCE checks will trigger under certain circumstances. cheers, Markus
