[Pkg-hpijs-devel] Bug#454117: hpijs uninstalls other printing systems

[Rogério Brito]

severity 490640 serious
thanks

Hi.

I'm seeing a disgusting situation here. I don't have any recommends
turned on (I need to keep my systems down to a minimum, without
recommends, suggests etc) and I'm using lprng to print.

Even trying to pull in the least amount of software (and constantly,
vigilantly looking for the packages that are installed on my system,
using security assessment tools, debfoster, deborphan etc), I have a
real problem here: just trying to install the hpijs driver for
ghostscript (which should not interfere with a well set-up system),
breaks unrelated software.

In particular, installing hpijs depends on cups-client, which conflicts
with (and removes) lprng.

On Jan 19 2009, Mark Purcell wrote:
> On Saturday 17 January 2009 19:58:25 Leif Hornsved wrote:
> > I've noticed a small problem with the latest change, version 2.8.6.b-4
> > of hplip. It now depends on libsane, 
> 
> Leif,
> 
> Thanks for your update to this bug report.
(...)
> > which means I get 14 MB of unnecessary packages.
> 
> Unfortunately the way hplip is packaged upstream it isn't trivial to separate 
> the scanning from the printing functionality. It maybe possible, by moving  
> /usr/lib/python-support/hplip/python2.5/scanext.so to it's own package, but 
> I'm certainly not going to attempt this so deep into the lenny freeze.

Can we have a solution here? This breakage of lprng is obviously
undesired, especially for devices with limited RAM (like embedded
devices that are meant to be used as a print server).

<digression>
BTW, I see this as a particular case of an undesirable change in Debian,
something that has been slowly creeping into our distribution.

Another notorious example of this problem is X depending on HAL, also
bringing in compulsorily the Multicast DNS things, a console font
setting to every system and many other dependency bloats out there. This
is quite an unfortunate situation.

Another situation of this is with the multimedia programs/music players.
</digression>

But, back to the present problem, just trying to get hpijs, we get cups,
which, in turn, grabs ssl-cert (huh? I just wanted a driver for
ghostscript, please), also brings in ttf-freefont (oh, just wanted a
driver for ghostscript!), foomatic-filters (again, just a single driver
for ghostscript, please), hplip (which, in turn, pulls in libsane, among
others), cupsddk (really?) :-(.

Also installed:

* framework for defining and tracking users
* a library for reading voltages of processors, fan speeds

And I just just wanted a driver for a language interpreter.

And what I get is my printing system broken. This can't be acceptable,
by any reasonable means.

> At one point the hpijs package was arranged for those with a printing only 
> requirement and hplip was aligned for both printing/ scanning.  But this isn't 
> possible at present (See Bug #454117)
> 
> > libsane is 10 MB. By breaking some
> > recommendations (not installing avahi-daemon etc) I can reduce it from
> > 14 extra packages to 4 packages and about 11 MB. I don't use a scanner
> > and I don't think I have any use of avahi-daemon.
> 
> Recommendations are certainly not necessary for any package functionality, 
> thus you can safely remove those packages.

See my comment above regarding the packages.

> However this still does mean you need to keep the libsane libs in the
> current package configuration.

And what about the main point of this e-mail, which is the breakage of
unrelated packages?

I remember that I could just install hpijs and work fine with
ghostscript and standard unix components.  It seems that we have lost
that ability.

I really think that the project should allow more flexibility of the
installation of the system.

The package manager front-ends already, by default, pull in "recommends"
and users, in general, won't loose anything with some dependencies
demoted to recommendations.


Sincerely yours, Rogério Brito.

P.S.: And the ironic thing is that the debconf prompt of foomatic asks
if the user wants to use lprng as a print spooler, which is something
that will be removed. This inconsistency is not something that one would
expect.

-- 
Rogério Brito : rbrito@{mackenzie,ime.usp}.br : GPG key 1024D/7C2CAEB8
http://www.ime.usp.br/~rbrito : http://meusite.mackenzie.com.br/rbrito
Projects: algorithms.berlios.de : lame.sf.net : vrms.alioth.debian.org