On Tue, Sep 9, 2008 at 6:03 PM, Eric Belanger
<[EMAIL PROTECTED]> wrote:
> On Tue, 9 Sep 2008, Anders Bergh wrote:
>
>> On Tue, Sep 9, 2008 at 3:46 PM, Ronald van Haren <[EMAIL PROTECTED]> wrote:
>>>
>>> so how would you know in that case if someone uploads a new package
>>> only for one architecture but the package is buildable for both
>>> architectures?
>>>
>>> Ronald
>>>
>>
>> Don't know, but TUs uploading new packages should try to build them on
>> 64-bit as well (if it is possible).
>
> I believe (and hope) that TU who have access to a x86_64 machine already
> provide x86_64 versions of their packages. BTW, gcarrier has a build machine
> for x86_64. Ask him for access if you don't have a x86_64 machine.

Yes, definitely. I meant that all TUs should have access to a x86_64
machine, which means all new community packages will have arch=(i686
x86_64) if they can be built on both architectures. If a TU only sets
arch=(i686) one should assume it doesn't work on x86_64.

>> Some of these packages will never
>> build on 64-bit, or be usable on 32-bit. Perhaps something like
>> arch=(i686 !x86_64) would be useful...
>
> In this case, !x86_64 will be taken as an achitecture. A comment after the
> arch field would be more aproppriate, e.g.:
> arch=(i686) # Binary blob, doesn't work on x86_64
>
> --
> This message has been scanned for viruses and
> dangerous content by MailScanner, and is
> believed to be clean.
>
>

Yes, it was a suggestion for how some packages (such as lib32-*) could
be ignored on that page.

-- 
Anders

Reply via email to