Re: Removal requests submitted for CERNLIB packages
Hi Lifeng,
On Fri, Mar 4, 2011 at 5:57 PM, Lifeng Sun <lifongsun@gmail.com> wrote:
> Hi Kevin,
>
> The FTBFS bug of CERNLIB is actually an issue of diffutils, when I
> built it after squeeze released last month, the bug disappeared, so I
> think this bug could be closed now.
>
> I would like to adopt the CERNLIB-related packages if allowed.
It's certainly fine by me, just be aware that I won't be able to
provide much (if any) mentoring support. You and Bastien will need to
find sponsors and/or join the Debian Science team for uploads. Do
please re-read [1] one more time ;-)
Looks like FTP-masters have been very responsive at following up the
removal requests, so when you have packages to upload, they'll need to
go back through NEW. Sorry for the hassle, I hadn't thought anyone
cared at all about these packages anymore based on the lack of
follow-up on any FTBFS bugs.
If you intend to keep the current Debian packaging, you'll need to get
the cernlib source package through NEW first, because the other
dependent packages use an infrastructure of included makefile
fragments shipped in the cernlib-base-dev package (if I remember the
binary package name correctly). See cernlib-xxxx/debian/cernlib.mk
and .../cernlib-debian.mk (again, if I remember the filenames
correctly).
Some miscellaneous comments follow that are of the "it would be
nice..." variety, but I don't have any personal interest in the
packages any longer so please do not take them at all seriously.
I'd suggest not expending much effort (if any) on the mclibs and
geant321 source packages. I'd expect they are less useful than the
other stuff (due to being largely superseded by Geant4 and Pythia8;
the latter is under GPL so someone might want to package it at some
point...)
On the other hand, the packaging might very well benefit from a
redo-from-scratch, as it's evolved very cruftily over the years. Part
of the point of the extra cruft was originally that it should be
possible for non-Debian users to be able to build easily from the
source package, but maybe that isn't so important these days? A lot
of the cruft is necessary though, as Cernlib's original source
building infrastructure is an utter abomination.
It could also be useful to update and make available the "Cernlib on
Debian FAQ" somewhere in the packages [2].
> BTW: I am preparing to package ROOT.
Wonderful! I hope that you are planning to start from the existing
packages at http://snapshot.debian.org/package/root-system/5.24.00-1/
(unless of course you are making a conscious decision to redo the
packaging from scratch) -- Christian put in a truly enormous amount of
work into the packaging, particularly in putting up with all my
nitpicks as his original sponsor ;-), and it's really unfortunate that
he ran out of free time to work on it and that it didn't succeed at
getting into the Squeeze release.
You should be aware that ROOT upstream is pretty impatient (at best)
with packaging-related issues [3,4] and (with some justification) not
convinced that distributor packaging is a good model for shipping ROOT
[5]. Athough I haven't heard from Christian lately, these may
possibly have been factors contributing to the recent lack of work on
the packages.
One issue that may have irritated upstream is the fact that Christian
felt strongly about keeping the Debian packaging stuff in upstream
revision control (so that users could run e.g. "make debian" to build
.debs during the period in which there were not officially provided
Debian packages). If he is no longer able to spend time on the
packages, that's a decision that a new set of maintainers could
conceivably revisit.
[1] http://lists.debian.org/debian-science/2008/09/msg00028.html
[2] http://www.jlab.org/~marki/cernlib_on_debian/cernlib/faq.html
[3] https://savannah.cern.ch/bugs/?func=detailitem&item_id=30944
[4] http://root.cern.ch/root/roottalk/roottalk10/0339.html
[5] http://root.cern.ch/root/roottalk/roottalk10/0335.html
best regards and good luck!
--
Kevin B. McCarty <kmccarty@gmail.com>
WWW: http://www.starplot.org/
GPG: public key ID 4F83C751
Reply to: