Re: Opteron support in dpkg
Alex Perry <alex.perry@qm.com> writes:
> I like John's summary (below). To add my 2c, as someone who has been
> putting in sweat on this port,
...
> "sparc" "i386" and similar names. For complete consistency, which was
> the asserted reason for the name change, we should use "opteron" as
> the architecture on that basis.
Opteron is the first alternative that wouldn't cause problems and
isn't braindead. - and _ in the name are not acceptable from a
technical point.
> "netbsd-x86-64" and encourage all the package maintainers to rewrite
Bsd uses amd64 in config.guess/sub.
> their parsing scripts that are trying to make sense of this mess for
> multiarch purposes.
>
>
> On Thu, Jun 03, 2004 at 05:36:41PM +0200, Goswin von Brederlow wrote:
>
>>Currently only the following are actively working on the pure64 port
>>(from the keyring that governs who may upload):
>>
>
> I didn't realize that you had to be a developer and put a key on the
> upload ring in order to be asked whether a decision was the correct
> one to take. John, would you please be my advocate and mentor ? It
> sounds like I _have_ to go through NM processing in order to get the
> opportunity to say "Listen to John!".
Actually anyone in the debian-amd64 can put his key into the
keyring. That your key is not in there just shows that you haven't
added a single package since the gpg check was turned on.
That means you haven't directly contributed to the pure64 archive. But
there is more to the debian-amd64 port and everyone is welcome to
comment and suggest things.
Since John says what we all want to hear lets listen to him. Very good
suggestion. .)
> Alex.
>
> John Goerzen wrote:
>
>>On Thu, Jun 03, 2004 at 11:23:13AM -0400, Stephen Frost wrote:
>>
>>
>>>Personally, my preference is to go with whatever produces:
>>>
>>>dpkg --print-architecture = amd64
>>>dpkg --print-gnu-architecture = x86-64 (or x86_64, whichever is right)
>>>
>>>
>>
>>I agree. Let's look at some of the arguments that have been raised
>>recently:
>>
>> * amd64 ties it to AMD, but we expect Intel to release a compatible
>> processor as well.
>>
>> This does not seem to have bothered us when dealing with i386, which
>> is Intel's chosen name, as opposed to K7 or others that have
>> developed extensions to Intel's specs. If AMD designed the spec,
>> I see no reason to avoid giving them credit just as we have done
>> with Intel or others. Besides, this is the chosen name for the
>> platform.
>>
>> The fact that AMD released i386-compatible CPUs didn't lead anyone
>> to suggest renaming i386 to x86 or ia32. In fact, the existance of
>> AMD's compatible CPUs predated the existance of Debian itself.
>>
>> * x86-64 corresponds better with GNU tools and the kernel.
>>
>> No, it doesn't. They use the underscore. The ensuing confusion
>> promises to be worse.
>>
>> Besides, GNU tools often say i486 for our i386 distro.
>>
>> * We could use the underscore later.
>>
>> That means switching our name twice more. A huge pain. Plus, we
>> don't even know if that is the case.
>>
>> * Others use x86-64.
>>
>> No, they use x86_64. Actually, plenty of others use amd64, including
>> Gentoo, NetBSD, FreeBSD, Mandrake, and some other smaller distros.
>>
>> * Renaming is easy.
>>
>> As Goswin has pointed out, it's not all that easy and in fact is
>> quite time-consuming.
>>
>>I am all for amd64.
------------------ :)
MfG
Goswin
Reply to: