[Pkg-cups-devel] Etch frozen, so what can we do?

[Kenshi Muto]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hello Debian CUPS guys,

As you know Etch is frozen yesterday.
I was working for 1.2.7-2 with newest upstream snapshot, but I'll
revert some of them. Because it's too hard to maintain two versions
at the same time, I'd like to concentrate to Etch and commit only for
RC, important or l10n bugs (as release managers described).

Let's see changelog of 1.2.7-2 I'd done.

  * Apply upstream r6149. Fixed some bugs around raster handling,
    printer status, IPv6 and so on.

Well, this changelog seems vague to understand.
See upstream changelog also.

CHANGES IN CUPS V1.2.8
       - Some types of Sun raster files did not print correctly
         (STR #2107)

minor?

       - Raw PBM files did not print correctly (STR #2106)

normal?

       - The SNMP backend no longer uses IPP with HP printers,
         as some recent firmware versions appear to not work
         (STR #2055)

minor because we don't use SNMP by default.

       - cupsMarkOptions() did not handle the
         multiple-document-handling option (STR #2135)

normal or important?

       - lpstat did not show the local job ID of active printers
         (STR #2125)

normal?

       - The backends incorrectly used STATUS:
         media-tray-empty-error messages for out-of-paper
         conditions (STR #2123, STR #2124)

important?

       - cupsGetPPD2() returned the wrong error when the PPD
         file did not exist (STR #2122)

normal or important?

       - cupsDoAuthentication() did not translate the password
         prompt (STR #2121)

minor but i18n issue.

       - httpGetLength2() did not handle error messages without
         content correctly (STR #2133)

normal or important?

       - Added support for 32/64-bit libraries on HP-UX Itanium
         systems (STR #2115)

can ignore.

       - Fixed a configure script problem with the 32/64-bit
         library support (STR #2114)

can ignore? (we use configure script of upstream tarball
instead of using autoconf)

       - The PostScript filter did not properly output document
         setup commands for reversed output (STR #2111)

normal or important?

       - The scheduler did not parse IPv6 netmasks properly (STR
         #2117)

important? (IPv6 is important issue for Etch)

OK, next changelog of 1.2.7-2.

  * Trying to revert seteuid way of upstream. (disabled 06_disable
   _backend_setuid patch to let backend works as upstream imagined)
    Put 66_setegid.dpatch to set GID.

This is just "trying" so it's OK to drop this.
But I'm still confusing about uid permission. Seteuid seems change
only UID so setegid lp is needed also, I think.
lp:root permission problem is reported sometimes.
mpitt, I believe you know well about it :) Does 06_disable_backend_setuid
work correctly without RunUser mode?

  * Add Printcap/PrintcapFormat/PrintcapGUI variables to cupsd.conf
    file. (closes: #401981)

Well, I think it seems serious level - FHS policy violation.
Printcap file (indicates /etc/printcap if we don't define) will
be made automatically by CUPS scheduler daemon.

Unfortunately current code of cups-bsd/cupsys postinst won't
replace /etc/printcap by symlink if /etc/printcap is real file.
So we have to add something workaround.

Conclusion:
- - Debian CUPS svn branches/cups-1.2 will be frozen also after
  cleaning :)
- - Splitting patches from upstream SVN is bit hard but it will be
  needed to follow the release basis.
- - We have to fix Printcap variable problem... includes modifying
  postinst.
- - lp:root problem exists. How can we solve?
- - Any other bugs/improvements we should fix for Etch?

I'd like to hear your comments :)

Thanks,
- -- 
Kenshi Muto
kmuto at debian.org
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.5 (GNU/Linux)
Comment: Processed by Mailcrypt 3.5.8+ <http://mailcrypt.sourceforge.net/>

iEYEARECAAYFAkV9/8UACgkQQKW+7XLQPLG0fwCgh+PZBZMxeuavNbW2taM/9c43
0csAn2n2PnL5wjF2q3yBWOhx+3RlgSSs
=TLfz
-----END PGP SIGNATURE-----