Planning the Fedora 11 Mass Rebuild

Panu Matilainen pmatilai at laiskiainen.org
Sat Feb 14 15:49:52 UTC 2009


On Fri, 13 Feb 2009, Tom Lane wrote:

> "Tom \"spot\" Callaway" <tcallawa at redhat.com> writes:
>> On 2009-02-13 at 15:16:17 -0500, Tom Lane <tgl at redhat.com> wrote:
>>> Hmm, I've got some specfiles that pay attention to the output of "uname -i".
>>> Should I expect that that might change from i386?  Is there a different
>>> more-recommended way to get the arch type in a build script?
>
>> Well, I don't think the uname -i will change, but the rpm target will
>> now be 586, so, they're not going to match anymore, where they did
>> before. You might be able to use %{_target_cpu} instead. It really will
>> depend on what value you're trying to get.
>
> I want to get "i386" or "x86_64", etc.  This is to be able to assign
> appropriate names to files that vary between 32- and 64-bit builds
> for multilib safety.

You might want %{_isa} then instead of %_target_cpu/%_arch and the like. 
%{_isa} for i386, i586 and the like is always x86-32 for x86-64 for x86_64 
family. Or if you just care about the number of bits %{__isa_bits} gives 
you the "bitness" of arch you're currently building for.

 	- Panu -




More information about the fedora-devel-list mailing list