[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Debian AMD64 Port Ready

On Sat, Jun 12, 2004 at 10:28:51AM +0200, Martin Schulze wrote:
> Chris Cheney wrote:
> > > > Sid. The port is currently at 97% compiled with most of the remaining
> > > > packages having FTBFS RC bugs filed for unrelated reasons. We have also
> > > > finished debian-installer for the AMD64 port and generate daily builds.
> > > > All that still remains to be done is for dpkg to include the amd64
> > > > patch, for archive space be given to the port, and for an official
> > > > buildd to be setup.
> > > 
> > > I also thought that the pure AMD64 port is technically ok, but
> > > worthless from a usability point of view since it is not compatible
> > > with AMD64 ports of other distributions.  If that's the case indeed,
> > > and if you are talking about the pure amd64 port, I don't believe
> > > Debian should include it in sid.
> > 
> > I have not heard of any compatibility issues, there is a lib64 dir just
> > as there is on the other distributions. It just happens to be a symlink.
> > Also, Multiarch is expected to be adopted by FHS/LSB and makes the
> > current distributions obsolete wrt their library placement in any case.
> > And since when is compatibility with other dists more important that
> > having support for an arch at all?  How is not being compatible with
> It has always been an issue.  Since AMD64 will probably be adopted
> commercially, there will be third party applications, that will only
> run on other distributions and not on Debian, which would make the
> Debian port pretty useless if you want to run only one other third
> party application which was compiled for another distribution.
> We've always tried very hard to keep our glibc compatible with the one
> from other distributions.

The /lib64/ld-linux-x86-64.so.2 is in the correct location and the glibc
is not modified in any major way that I know of so it should be
compatible with the LSB. Debian has never been more than trivially
compatible with other dists since our regular libraries have never been
in sync with the other dists. A lot of that has to do with the fact that
we release every 2-3 years when other dists release every 6 months.

> > other distributions useless usability wise? I have never had the reason
> > to install another distributions rpm package on my system and have not
> > heard of anyone else needing to either, on amd64 or even i386 for that
> > matter. Of course Debian will be useless usability wise if sarge does
> I acknowledge that this is your personal experience.  It does not mean
> that it is everybody's experience.  Especially, it does not have to
> imply that no user will ever want to run third party software.

True, and since our ld is located in the right spot for AMD64 LSB
compatibility and we have ia32-libs like ia64 we can support i386
binaries just like everyone else as well. Of course if someone needs
extreme compatibility with another dist and not just LSB compatibility
they will need to use that other dist as they do today instead of using
Debian i386.

> > not ship with AMD64 arch support since by 2007 (sarge+1) all desktop
> > systems will be AMD64/EM64T/IA32E based. BTW Intel is expected to enable
> > the 64bit extensions on the P4 3.2GHz+ chips by the end of August 2004.
> Since the amd64 CPU is fine running i386 code, how could the i386 port
> be useless?

The i386 port is much slower than running amd64 code, on average 20-30%
slower and yes this has been tested and proved. Some people also
actually need the 64bit support that the arch provides. Debian has
already lost a lot of users since AMD64 was released in March 2003
because every other major dist supports it now, some for over a year. So
it is totally unacceptable for Debian to suggest to users to use the
i386 port instead and Debian will continue to lose many more users if it
sticks to that line. Also, since Debian already has a fully working
AMD64 port it would be very stupid not to release Sarge with it.


Attachment: signature.asc
Description: Digital signature

Reply to: