Re: Sarge may be last Debian release for 32 bit sparc systems
On Tue, Jul 26, 2005 at 09:11:45AM +0200, mag wrote:
> 2005-07-25, h keltez?ssel 14.06-kor Blars Blarson ezt ?rta:
> > In my opinion, we should drop support of all 32-bit sparc systems from
> > Etch due to lack of people willing to spend the time to support them.
>
> I have just decided to give a try to an old SS10 :(
> Maybe it is another reason to actually begin it ;)
>
> Nonetheless I understand the reasons for the proposal, and cannot offer
> more support than trying to answer questions related to areas where I
> will be able to overcome problems.
>
> Is it true that only kernel and closely related packages should mean
> special problems for sparc32?
If there's no kernel, the userspace doesn't matter. While it's hard to
imagine, these machines were newish around the time of the 2-digit MHz
clock speed Pentiums, before MMX even. Well, we still support Pentium
machines, so why not sun4m machines? The sad answer is both that PC
hardware is more prevelent by far, and that it hasn't changed that much
since then. When I say it hasn't changed that much, what I mean is that
it has massive backward compatibility by comparison. That's actually
bad, which is why it's sad. sun4m hardware is vastly different from
US2 hardware, and not just in the 32 v. 64 way. And even inside that
class of machines, there are a lot of incompatible variations.
The truth of the situation is that the massive technological beating
the x86 industry has delivered on modern SPARC processors has dealt
that architecture a mortal blow, but it's still thrashing. But it has
killed the pre-ultraSPARC hardware cold dead at this point. All that is
keeping it supported up to now is sentimental fools. I count myself in
there somewhere. The dang hardware is so slow I run my SS20 with NFS
root, because it's way faster than the speed of a local disk! I load
the kernel off the disk and then I have a little rc script that spins
the disk down so I don't have to listen to that banshee. I mean, hey.
> I would be interested in an estimate of the sparc32 userbase size.
Still one here, but I admit that it gets turned on a lot less than my U2.
And I have to admit that I'm considering pulling the plug on it, laquering
it and putting it in the museum for good. Do you want to spend weeks
fixing the kernel problems for me? Before you start, you should know
that it's a dual Ross HyperSPARC machine, much different from a dual
SuperSPARC. Sigh.
a
Reply to: