Re: Breaking /emul/ia32-linux for squeeze
md@Linux.IT (Marco d'Itri) writes:
> On Mar 16, Simon Richter <email@example.com> wrote:
>> Well, it would get i386/amd64 in line with sparc/sparc64, powerpc/powerpc64
>> and s390/s390x. That would allow us to get rid of a lot of specianl cases,
>> including the hack for libc6-386.
I don't see sparc/sparc64, powerpc/powerpc64, mips/mips64/mipsn32,
mipsel/mips64el/mipsn32el sparc/sparc64, powerpc/powerpc64, s390/s390x
or arm/armel/armebai as having solved the multiarch problem.
Except maybe s390x as the s390 port seems to want to drop 31 bit
support and go completly 64bit.
> I think it would be very helpful if somebody could summarize why a
> multiarch system is useful, except for the obvious case of installing
> proprietary i386 software on amd64 systems.
So I can have a 64bit mysql or postgres for my huge collection of
porn on my mipsel system.
So I can install a 32bit gcc on my amd64 as that eats half the ram
compared to the 64bit one while keeping a 64bit gzip as that runs 20%
So I can install wine on my amd64 without the maintainer having to
include a prebuild wine in the wine source.
So I can get rid of all the biarch packages that are quite ugly and
only work 75% right.
So I can install a linux iceweasel and flash plugin on freebsd.
So I can install and compile against 32bit uclibc packages on my amd64
and use the results on my i386 based netapp fileserver.
And then it should cover cross-compiling as well for all those that
want to build packages for their embedded arm boxes.
And as an extra gimmick it would allow upgrading a i386 lenny to amd64
Note that 7345 out of 16030 (46%) reported amd64 systems have
ia32-libs installed. 4453 (28%) also have ia32-libs-gtk installed.
So even just "proprietary i386 software on amd64" is a huge chunk of
debian user. And those numbers will more likely go up than down as
i386 users overcome their fear of updating to amd64.