On Wed, Oct 27, 2010 at 11:42 AM, Martin Pitt <email address hidden> wrote:
> Matthias Klose [2010-10-27 14:44 -0000]:
>> I don't get it. I certainly prefer a working package on all supported
>> platforms, even if it runs slower on some of them. We would do the same on
>> other platforms if we would detect a package built with, e.g. -msse2
>
> I agree. We specifically said that we don't want NEON enabled in any
> package (yet), so this has never been intended.
Sure, but as it does improve the performance a lot on hardwares that
support NEON, I also agree that this is post release regression.
For users for sure it would sound weird to have a fix that actually
decreases a lot the performance of qt based applications on their
hardware after an update.
On Wed, Oct 27, 2010 at 11:42 AM, Martin Pitt <email address hidden> wrote:
> Matthias Klose [2010-10-27 14:44 -0000]:
>> I don't get it. I certainly prefer a working package on all supported
>> platforms, even if it runs slower on some of them. We would do the same on
>> other platforms if we would detect a package built with, e.g. -msse2
>
> I agree. We specifically said that we don't want NEON enabled in any
> package (yet), so this has never been intended.
Sure, but as it does improve the performance a lot on hardwares that
support NEON, I also agree that this is post release regression.
For users for sure it would sound weird to have a fix that actually
decreases a lot the performance of qt based applications on their
hardware after an update.