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

Re: is m68k really ready for testing?

[re-sending because I hit the wrong button -- again]

On Sat, Apr 05, 2003 at 03:26:01PM +0200, Marco d'Itri wrote:
> Today, with the help of some m68k porters, I investigated in the crest
> unstable chroot why mutt does not build anymore on this architecture,
> keeping out of testing an important security fix (possible remote
> exploit).
> aj announced on the d-i list that he wants to do a sarge preview by
> end of the month, but with so many packages broken I really wonder if
> it's still appropriate to consider m68k in a releasable state,

Currently? No. But does that matter? We're not even close to releasing:
debian-installer doesn't work on a lot more than ia64 and i386, the
g++-3.2 transition isn't quite finished, and we still have a whole lot
of RC bugs waiting for fixes.

The problems you mention aren't all unknown (some are, and will be
fixed once our backlog is cleared), but all of them are a result of
problems with the toolchain. These problems have now been fixed, thanks
to the great work by Matthias Klose and others.

> given the
> damage to other architectures.

What damage exactly are you referring to? Let me spell it out for you
again: not a single architecture can do damage to another one, unless
I'd start uploading m68k binaries in a package for powerpc, or so.

Problems such as the one m68k is suffering from now can happen to _any_
architecture, and in fact, have happened to other architectures before.
We've had some hardware problems during the last week; as a result, half
of our autobuilder capacity had been out. The resulting backlog has now
almost been cleared...

Yes, there have been problems with our toolchain. Does that mean we're
suddenly showing the world that we can't keep up, and that we should be
dropped? If someday there are problems with the s390 toolchain, will
you get up and scream that we should stop support s390, because
"nobody's using s390 anyway"[1]?

Do remember that unstable is there to be broken from time to time. As
long as bugs get fixed (and they are getting fixed), nothing's going

> I hope that the m68k porters will not have bad feelings about this:
> I have nothing against their toy, as long as it does not get in the
> of other architectures which are actually used by people.

m68k is also 'actually' used by people; if not, we wouldn't be here.
Would you please leave the choice on whether something is valid or not
to our users?

> (Please do not suggest "fix these bugs yourself". I looked at my own
> package and determined that the bug is not my fault, I have no time to
> dedicate to a doorstop architecture.)

If you want us not to have bad feelings about anything, I'd suggest you
keep the insults for yourself.

Ignoring that, I cannot help but notice that there are other viable ways
to fix this particular issue. Compiling your package does not fail on
the actual compilation; it fails on converting the documentation from
SGML to another format.

It has been suggested that you split off your documentation to an
architecture-independent package. Besides fixing this particular issue,
such an action would also, approximately, reduce the size of your
architecture-dependent packages by half a megabyte. Since
architecture-dependent packages are reproduced for all our
architectures, that would reduce the usage of your packages on our
mirrors by a few megabytes. You refused, because you didn't want to 'add
another micropackage'. Although questionable, I can understand this

Another possibility would be to ask aj to force this particular update
into testing. Since it concerns a security update, it is not unthinkable
that aj would comply.

Instead, you chose to question the validity of Debian/m68k. One could
wonder how long Debian would survive if we'd start dropping
architectures that are suffering from temporary problems.

Not long, I'm afraid.

[1] Note to s390 porters(and others): that's all hypothetical, of
wouter at grep dot be
"An expert can usually spot the difference between a fake charge and a
full one, but there are plenty of dead experts." 
  -- National Geographic Channel, in a documentary about large African beasts.

Attachment: signature.asc
Description: Dit berichtdeel is digitaal gesigneerd

Reply to: