Okay I ran across an interesting issue trying to get scipy compiled with atlas support. lapack and atlas both provide liblapack.so.3, in different locations, but accessible to the linker. scipy builds against atlas just fine, and runs on systems just fine...unless lapack is installed on the system. Once lapack is installed the link prefers the lapack package's version of the same library and runtime errors occur. I think I can work around this in scipy packaging... but the question is, is it a general problem that lapack and atlas are providing link incompatible versions of the same library? The packages which reqoquery --alldeps lists for atlas and lapack have a significant overlap. Are all those overlapping applications affected by this runtime linking problem? -jef -- fedora-devel-list mailing list fedora-devel-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/fedora-devel-list
