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

Re: Uninstallable Packages

On Tue, Oct 05, 1999 at 10:13:51PM +1000, Anthony Towns wrote:
> I'm experimenting with a script to work out whether packages are
> installable or not. I figured the world at large might be interested in
> some of the results.

Some more experimentation. This time limited to main, but across i386,
m68k, sparc, alpha and powerpc architectures. This doesn't reflect
whether packages in an architecture are up to date or not.

Statistics are:
	i386 has 55 uninstallable packages -- 3494 total
	alpha has 73 uninstallable packages -- 3246 total
	sparc has 87 uninstallable packages -- 3191 total
	m68k has 92 uninstallable packages -- 3205 total
	powerpc has 100 uninstallable packages -- 3149 total

There are 201 distinct uninstallable packages all up, 3561 in total.

For interest's sake:
	arm has 568 uninstallable packages -- 2003 total
	hurd-i386 has 398 uninstallable packages -- 1033 total

Note that these are largely unverified.

I'm working on the basis that every package that can be built for an
architecture (ie, has Architecture: any, all, or includes whichever
architecture specifically) should be installable on that architecture.

Is that true?

libtricks gives me pause to wonder:
	libtricks(1943) OKAY: [m68k, sparc, powerpc] BAD: [i386, alpha]

It's bad on i386 and alpha because libc6 conflicts with it. IIRC, it
got a little too friendly with libc's internal structure and hence died
when glibc2.1 came out. I'm not sure why sparc and powerpc don't also
have problems, though.

OTOH, m68k is still glibc2.0 based, so libtricks is fine there. How should
this be handled? By adding an `Architecture: m68k' ? But that's not very
accurate, it compiles fine on i386 if you've got glibc2.0. But on the
other hand, it doesn't seem like it should be included in Debian for
i386 or alpha because it's perfectly useless.

In this case, we've got an easy out, fakeroot works fine everywhere:
	fakeroot(676) OKAY: [i386, m68k, sparc, alpha, powerpc]

But I wonder if, generally, it's reasonable to require all architectures
to make all packages available (neglecting minor things like packages with
i386-assembler, or whatever).

I'd like to think it is, but...

The list of which packages aren't installable on which architectures is at
. It's 11kB or so, but probably not too exciting.


Anthony Towns <aj@humbug.org.au> <http://azure.humbug.org.au/~aj/>
I don't speak for anyone save myself. PGP encrypted mail preferred.

 ``The thing is: trying to be too generic is EVIL. It's stupid, it 
        results in slower code, and it results in more bugs.''
                                        -- Linus Torvalds

Attachment: pgpxVeAnnmmg0.pgp
Description: PGP signature

Reply to: