On Wed, 2012-04-18 at 15:37 +0200, Phil Knirsch wrote: > Yea, it's kinda butt ugly, but the problem is that rpm doesn't have an > API for it, even less so for the python part that would be required.
Yeh, just seems that one of the bester cases here would be having an "rpm API" that says "give me the current AT_PLATFORM and/or AT_*". Even if we don't move/merge "all" of the arch detection code into rpm, at least having those would be good. > The latest patch version for rpm at least does it in a decently sane way > where a struct is being populated with AT_PLATFORM and AT_HWCAPS > information to be used for defaultMachine() stuff. > > The other 2 options are to either look at /proc/cpuinfo or to use ENV > where at the end the AT stuff is being "injected", but the IBM folks > strongly recommended not to use the former Do they say why they don't want us looking in cpuinfo? We do that on most arches anyway ... I just know anaconda/mock/pungi/something is going to barf when we have to start running other programs. _______________________________________________ Yum-devel mailing list [email protected] http://lists.baseurl.org/mailman/listinfo/yum-devel
