Jonas Smedegaard wrote:
By Debian social norms, Masayuki Hatta has to accept it. He is a quiet
guy, however, and these mails are cc'ed the package, reaching all team
members. So I would say that the lack of response, keyed with earlier
approval of moving to Git in the collab-maint group, can be interpreted
as implicit approval.
So please, let's move on :-)
OK.
Is unstable in a release freeze currently?
Nope. But the term "unstable" in "Debian unstable" refer to the distro,
not each package, and library changes should be treated with care at all
times (something I have learned the hard and embarrasing way with libgd
and uw-imap).
The split is not a library change. the binary packages libgs8 and
libgs-dev do not change by this step.
The only changes in the libraries are all the cherry-picked bug fixes,
but none of them changes the API. They only improve the stability and
the correctness of the output.
Yes, I understood that from changelog entries. Thanks for confirming.
My point above is that packages have been released that contained those
symbols related to libcairo linkage, and in principle there is now a
risk of other packages relying on those symbols and thus need rebuilding
and/or patches.
the libcairo symbols was one example - other symbols have disappeared
too since the release of Lenny (I have not kept symbols earlier than
that).
Also, concretely for the libcairo linkage, we should perhaps reconsider
if pulling in X11 libraries is still evil: With the improved X11
packaging it might no longer be too much of a burden.
To avoid pulling X11 is important for using head-less servers as print
servers. They need to run Ghostscript but one wants to avoid installing
X libraries only to satisfy Ghostscript's dependencies.
The libgd package has similarly been provided both with and without X11
linkage to optionally avoid bloat, but I expect to drop that now, as the
bloat is no longer as large as a few years ago.
As the X itself is already split, I would like to leave it this way. The
Cairo interface is still experimental AFAIK. So before doing changes
here I would like to know if the Cairo interface is already considered
stable.
Right. Checking only binary dependencies as I did can cause false
positives like that. But the potential list is longer: there's a bunch
of packages depending only on ghostscript-x even if at least one of them
() is known to link against libgs8.
Can you tell which packages?
Excellent. I notice now that you are also member of collab-maint, so
you already have write access to the ghostscript Git :-D
git clone ssh://till-gu...@git.debian.org/git/collab-maint/ghostscript
Feel free to commit changes directly there. Except for adding new
binary packages - please postpone such changes until after we've
released to unstable.
Oh, and if you are unfamiliar with some of the advanced CDBS features
then try read the README.source file and if still puzzled (e.g. about
the content of control.in and rules files) then please ask.
OK.
Till
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
