https://bugzilla.redhat.com/show_bug.cgi?id=820544 --- Comment #43 from Mads Kiilerich <mads@xxxxxxxxxxxxx> --- (In reply to comment #39) > Ooops! This is completely wrong for i686. Package resolvers will fail trying > to download x86_64 packages which are not available for i686. One would > first need to rebuild the src.rpm, which is highly recommended anyway, > because there might be conditional (arch-dependent!) BuildRequires in the > spec file, which only become active build requirements if the src.rpm is > built on the correct target arch. As you point out: The src.rpm dependencies (and thus yum-builddep) are not reliable anyway. It is for example (usually) not possible to do 'yum-builddep grub2' on i686. (IMO it would be better to deprecate yum-builddep and provide a simple way to install build dependencies from rpmbuild/fedpkg, using locally and correctly exapanded macros.) It do however IMO matter that build requirements are 100% reliable that if just the stated requirements are met somehow then we can trust that the build will succeed and be valid. A unspecific requirement can be ambiguous and fatal on multilib systems. IMO it is thus better to make the arch requirement explicit when it matters. But right, this is not how we do in Fedora where multilib has several grey areas ... and this is thus not the rigth place to discuss it. -- You are receiving this mail because: You are on the CC list for the bug. _______________________________________________ package-review mailing list package-review@xxxxxxxxxxxxxxxxxxxxxxx https://admin.fedoraproject.org/mailman/listinfo/package-review