x86_32 yum $basearch

[Panu Matilainen]

On Fri, 20 Feb 2009, Josh Boyer wrote:

> On Fri, Feb 20, 2009 at 09:20:26AM -0500, seth vidal wrote:
>> On Fri, 2009-02-20 at 14:12 +0000, Paul Howarth wrote:
>>> seth vidal wrote:
>>>> On Fri, 2009-02-20 at 09:42 +0000, Paul Howarth wrote:
>>>>> Now that Fedora 11 x86_32 is going to be based on i586 packages rather
>>>>> than i386 packages, does it follow that yum's $basearch will change from
>>>>> i386 to i586 and hence repository directory layouts changing too, or
>>>>> will it stay at i386?
>>>>>
>>>>
>>>> I don't think it will change. For a few reasons:
>>>>
>>>> 1. non-fedora repos
>>>> 2. non-fedora distros
>>>> 3. this is why we also have $arch in addition to $basearch
>>>>
>>>> make sense?
>>>
>>> Certainly from the point of view of being the least disruptive option;
>>> it'll just look a bit strange having i386 "Everything" repos with no
>>> i386 packages in them...
>>
>> Which is why we shouldn't have i386 repos. We should switch
>> fedora-release to using $arch and an i586 repo which is a symlink to the
>> i686 repo. Then when we phase out i586 in F12, we just remove the i586
>> symlink. That way, if anyone wants to maintain a build of fedora for
>> i586 or lesser machines then they can do so in the i586 path using
>> $arch.
>
> Or you could just have:
>
> x86
> x86_64
> ppc
> ppc64
> sparc
> sparc64

...or you could use rpm's idea of ISA to avoid yet another set of 
different names :)

Available to python with rpm.expandMacro("%{__isa}"), expands to things 
like:
x86-64
x86-32
ppc-32
ppc-64
..etc


 	- Panu -