[Pkg-cups-devel] Bug#681851: escputil: requires module usblp which is blacklisted by CUPS

[Till Kamppeter]

On 07/24/2012 09:09 AM, Didier 'OdyX' Raboud wrote:
> Now for Wheezy, I can't have a solid enough on the changes introduced after
> 1.5.3-1. 1.5.3-2 didn't migrate because the mips build failed; I suspect this
> can be solved by a build on (currently down) ball.debian.org.
>
> Till: are there atomic changes done since 1.5.3-1 that would be worth
> integrating into Wheezy? I'm willing to prepare a 1.5.3-2+wheezy0 that would
> revert 1.5.3-2 and add the smallest amount of changes towards wheezy if that
> helps giving Wheezy a better cups. (Keeping in mind that the Release Team will
> review the diff between 1.5.3-1 and whatever we want in wheezy.)
>

The changes are the following:

1. Addition of the ipp14 backend: The ipp14 backend is nothing else that 
the IPP backend of CUPS 1.4.x, which is code which has been in Debian 
for longer time already. I have re-added it but to not drop the new 
backend of 1.5.x with its new fixes and better standards compliance, I 
have added the old code as ipp14 backend. This backend will never be 
used automatically so regressions are not possible, but for the case of 
the new backend causing problems the user can switch to ipp14 manually. 
Changes are debian/patches/add-ipp-backend-of-cups-1.4.patch with the 
code and the Makefile entries and addition of an "ipp14" entry in Debconf.

This avoids regressions by incompatibilities of the new ipp backend with 
some devices (especially LiveBox router).

2. Improvements of the USB backend: These improvements fix several 
compatibility problems of the new libusb-based USB CUPS backend, 
especially it adds support for uni-directional devices, quirk management 
for devices with incompatibilities, and more. All changes are in 
debian/patches/usb-backend-further-enhancements.patch. Important here is 
to take the patch of at least cups 1.5.3-4 to have all fixes. It is also 
not possible to separate the fixes as the bugs got fixed by combinations 
of them.

This fix solves many problems of the original libusb-based USB backend, 
especially one does not need to blacklist the usblp kernel module to 
work around some of these problems.

3. Blacklisting of the usblp kernel module: In CUPS 1.5.3 the USB kernel 
module is blacklisted to work around several problems of the 
libusb-based USB CUPS backend. This causes other problems, especially 
ink level checkers like escputil and ink do not work, but also some CUPS 
backends coming with proprietary drivers from printer manufacturers do 
not work.

In 1.5.3-3 I have lifted the blacklisting again as with the fixes of the 
USB backend (2) the blacklisting is not needed any more.

All these changes got accepted as an SRU (Stable Release Update) for 
Ubuntu Precise and as the package got released for testing 5 of the 6 
reported bugs got verified as fixed witb these changes by their 
reporters, no regressions caused by these changes got reported.

I recommend to take them all into Wheezy as they fix a lot of bugs.

> Finally, #640939 should probably be fixed in Wheezy and we should avoid
> letting #682426 enter Wheezy either.

Bug #640939 (http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=640939) is 
not affected by the fixes mentioned above. Important for the CUPS 
package is that all calls of lpstat/lpinfo/lpadmin/lpr/lp/... in the 
maintainer scripts have to be done with "-h /var/run/cups/cups.sock" to 
assure that they never access a remote print server, even if there is a 
client.conf file pointing to a remote server. Default cupsd.conf needs 
to "Listen /var/run/cups/cups.sock" line, even if there are other Listen 
or Port lines to let CUPS listen to requests from the network. This 
should be already the case, please check.

Printer driver packages should always use the CUPS trigger and never 
their own PPD file updaters in their maintainer scripts, please check.

In Ubuntu we have all this in place and do not get such bug reports.

Bug #682426 (http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=682426) is 
not a CUPS bug. It needs to get reassigned to Ghostscript. The user 
needs to be asked to attach the input file and the bug needds to get 
forwarded upstream to Ghostscript. As workaround adjustments in 
cups-filters can be applied. In CUPS nothing can be done about it.

    Till