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

c2a transition status

It really is going pretty well.  I see no reason to think that there will
be any difficulty in finishing both the c2 and the c2a transitions completely
in etch.

At this point arm is catching up with its build backlog, but hppa is
dropping fast; HPPA is now doing worse than ARM.  HPPA may need to be
added to the list of architectures where the uninstallability count can
increase, since there's no progress on some of the nasty bugs like the
illegal instruction in glibc preventing ARTS from building.  If the HPPA
problem is dealt with one way or another, I think the KDE cluster for the
c2a transition (the largest cluster) will be a reasonably smooth hint.

It looks like the following need rebuilds (or removal from testing)
for the GNOME C++ cluster to enter testing:
* timfx
* libbuffy
* mysql-query-browser
* patchage
* quickplot
* tagcolledit (after libtagcoll-dev is ready everywhere, as noted by Enrico Zini)
Then it looks like that cluster will go in fairly smoothly with this hint:
hint libsigc++2.0/2.0.16-2 glibmm2.4/2.8.2-2 gconfmm2.6/2.10.0-3

Most of the other clusters are tangled up with kde in one way or another
(kword, abiword, goffice, and librsvg2 all indirectly depend on libgsf,
which makes an unfortunate tangle).  KDE unfortunately is one of the
worst offenders with regard to recursive library dependencies, so some
of this is probably unnecessary.

c2a *libraries* probably needing requeues:

* digikam -- should probably be requeued on sparc, powerpc, mips, mipsel,
  ia64, alpha; it didn't build because it needed new arts, which is now
  present (except for hppa).

* libaqbanking -- should be requeued on ia64 and hppa; failed to build
  because kdelibs wasn't ready,  which it is now.

* mysql++ -- FTBFS on sparc, looks like a transitory buildd problem -- requeue?

These packages have old versions in testing and can't get the 'c2a' versions
in because of build problems:

* boost -- FTBFS on hppa, bug 341174
* cppunit -- FTBFS on hppa, bug 341675, FTBFS on m68k
* dar -- FTBFS on ia64 (binutils bug, #34277),
         FTBFS on mips, mipsel (#342778)
* tse3 -- FTBFS on alpha, ia64 (#337820), not yet built on m68k, s390
* usrp -- FTBFS on ia64 (#342780), not yet built on arm, m68k
* eris -- didn't build on hppa because of atlas-cpp FTBFS on hppa

The following library packages still need renaming for the
C++ allocator change, and the old versions are in testing.
I've also excluded those which are already hinted to be removed
from testing, and I've done some dependency analysis: the full list
of recursive reverse dependencies for the source packages is present,
should you find it handy.

* libprinterconf -- in NEW queue for 2 days
* wftk -- in NEW queue for 3 days

* gnuift -- no reverse dependencies
* mercator -- cyphesis-cpp
* zipios++ -- enigma
* libmusicbrainz-2.0 -- quodlibet-plugins
   Obsolete library version, being removed
* xalan -- zope-zms, metapackage med-cms from debian-med
   Neither depend on the library directly so neither will need a rebuild.
   of its own (metapackage med-cms from debian-med).
* rlog -- encfs, libpam-encfs, gmailfs (Recommends)

One more c2a oddity which needs an updated version and has the old
version in testing:
* festival -- not a dynamic library, but there is a -dev package.
              Should be queued for rebuild to fix static library

Nathanael Nerode  <neroden@twcny.rr.com>

"(Instead, we front-load the flamewars and grudges in
the interest of efficiency.)" --Steve Lanagasek,

Reply to: