F12 to require "i686", but which CPUs do not qualify?

[Andrew Haley]

Kevin Kofler wrote:
> Joachim wrote:
>> I do not understand then, that there exist i686 packages which have
>> higher requirements.
> 
> Those packages need to be fixed.
> 
> I know there are some audio production packages which are building with SSE 
> enabled (and required, those packages don't do runtime detection), IIRC in 
> both Fedora and RPM Fusion, in blatant violation of the guidelines, and the 
> packager(s) refuse(s) to fix this (they even do it intentionally for new 
> packages, despite my objections in the reviews). If I'm not mistaken, most 
> of the offenders are owned by oget (Orcan Ogetbil), but if I were you, I'd 
> check all the audio production packages.
> 
>> Look at the ATLAS library for which I had filed a bug because only
>> SSE/SSE2/SSE3 variants are provided
> 
> This one needs to get fixed too, of course.
> 
> I've looked at how Debian is handling this, but they're stuck at an old 
> version (3.6.0), maybe exactly because of this issue. :-(
> 
> We need to provide "architectural defaults" for plain i686, even crappy 
> ones, they just need to work at all.

I think there's a valid case for making an exception to this: when a
package is an accelerated version of a particular library.  That is,
when the basic functionality of a library is available in a i686
Fedora package, but a special SSEx version of the library makes use of
faster instructions.

Andrew.