https://bugzilla.redhat.com/show_bug.cgi?id=820544 --- Comment #44 from Michael Schwendt <mschwendt@xxxxxxxxx> --- > The src.rpm dependencies (and thus yum-builddep) are not reliable anyway. Remains the question whether that is sufficient reason to break "rpm -qpR …src.rpm" queries and tools such as yum-builddep? Perhaps the RPM devs would also like to kill src.rpm Requires that are created from the BuildRequires. > A unspecific requirement can be ambiguous and fatal on multilib systems. ... for cross-arch building only (such as i686 on x86_64 *without* using Mock), and mainly because not all -devel subpackages contain arch-specific base package dependencies yet (despite the added entry in the Packaging Guidelines). The bad effects are not limited to yum-builddep, however. A manual "yum install 'gnutls-devel(x86-32)'" currently still pulls in a mix of i686/x86_64 packages, for example. And that's just one example. One doesn't even get the minimum build environment via arch-specific dependencies alone. No glibc-devel.i686 and similar packages. > ... and this is thus not the rigth place to discuss it. :) That would be something for the FPC. It would be half-hearted anyway that arch-specific base pkg deps are a MUST in the guidelines, but packagers don't update their spec files accordingly or even refuse to do so because they disagree with the guidelines. Not even glibc-devel contains an arch-specific base pkg dep yet. > multilib has several grey areas This topic is not limited to "multilib". I understand the intention to add %_isa to BuildRequires, but it's of questionable benefit and not free of problems either. -- 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