Re: Policy, updmap --enable and updmap.cfg in /etc or /var

[Florent Rougon <f.rougon@free.fr>]

Hi,

Florent Rougon <f.rougon@free.fr> wrote:

> Well, well, well. I looked at ucf this morning, and it is supposed to do
> just what I need. Unfortunately, my tests with --three-way all failed
> miserably (the file was never put into the cache), and since the docs
> don't say much about it, I had to read the ucf script and add debugging
> statements here and there to find out why it didn't work.

[...]

> As a consequence, I'm still pondering what I'm going to do about this
> issue...

I reported the aforementioned bug last week and Manoj fixed it quickly,
so I decided to eventually try ucf for lmodern (I also read[1] in the
debian-devel list archives that the current ucf could be considered a
prototype, which somehow confirms the judgement I stated in the previous
mail). I like the concept of ucf, and being able to update
/etc/texmf/updmap.d/10lmodern.cfg is a good thing.

So, I uploaded and tested version 0.92-4 of lmodern using ucf here:

  deb http://people.via.ecp.fr/~flo/debian sid/binary-all/
  deb-src http://people.via.ecp.fr/~flo/debian sid/source/

There are a few problems caused by ucf bugs that I just reported
(cf. #279259, #279261 and #279262), but I think that the benefits from
using it outweigh the drawbacks. So, unless I here your protests, I will
probably ask Sébastien Bacher soon if he can sponsor the new package (he
is my usual sponsor).

The main problem that I have noticed is that when upgrading from a
pre-ucf lmodern (i.e., <= 0.92-3) to the new package, if the admin had
previously deleted /etc/texmf/updmap.d/10lmodern.cfg, ucf will happily
recreate it when it is run from lmodern.postinst, destroying the user
configuration (that is, the deletion of the file). Fortunately, that
does not happen when the file has already been registered with ucf, so
if you install this 0.92-4 package, delete the configuration file and
then upgrade to a later version, ucf will not recreate the file (but it
will print a spurious error message, grrr...).

I tested all the scenarios I could think of, so it should be safe apart
from the preceding problem.


[1] http://lists.debian.org/debian-devel/2003/04/msg01441.html

    Manoj:

      "Using debconf is on the TODO list for ucf, and perhaps a rewrite
       of the current prototype in C for speed later down the line."


For Frank, from one of your last messages:

> Hope so, too. I'm even quite confident it will. Actually I am living and
> working in Zürich since Oct 1st, sleeping in a small chamber and sharing
> kitchen and bathroom with a student. My wife and all the stuff will move
                                       ^^^^^^^^^^^^^^^^^^^^^^^^^
> on Thursday, and we have a couple of friends to help.

Doh! I hope she doesn't read -tetex-maint! ;-)

-- 
Florent