Extremadura-Meeting: Report from Andreas Tille
Hi,
I just want to give a status report of my work at CDD / Debian-Edu meeting
in Merida last week. First of all thanks to the organisers for these
successful days. While sometimes forced to work ofline this sometimes
has the positive effect to work undisturbed by external influences. Moreover
the luxurious hosting, the delicious meals and the great weather conditions
overruled the weak connection. Morover we had some kind of mysterious
entertainment (see below). ;-)
1. cdd-dev
- Jonas suggested to implement DebTags into creation of debian/control
with which makes specifying dependencies more flexible. The original
plan was to create a debian/xcontrol file which is processed afterwards
by debian-xcontrol which "sanitizes" dependency lists. While
debian-xcontrol does not yet have the ability to process DebTags
it should be implemented for the profit of other non-CDD related
packages.
Thinking twice about this plan I decided that it should be perfectly
doable to include this functionality into the new cdd-gen-control
tool that has to be rewritten anyway. Just adding those dependencies
that are in DebTags in addition to the dependencies that are hardcoded
in the tasks file should be cheap.
- Teaching others how to use cdd-dev: Well, I've got the hint that
some important points are missing in the docs which definitely
should be enhanced. I managed to include two major points but
there is much more to do to make a _fine_ manual out of the existing
one.
Most changes in
http://people.debian.org/~tille/cdd/
happened in Attachment A and B.
2. cdd-webtool
- The original plan was a rewrite of the current code for Debian-Med
stuff. While I adjusted the scripts that are producing the pages
which I call "WebSentinel" of our meta packages to build the packages
for other CDDs as well it needs several hacks and a quite tricky
templating system that should be replaced. The current system seems
to be even worse once it comes to translations.
Jonas gave me the hint to python-genshi which seems in fact a cute
templating system that might perfectly fulfill all needs for the
task to do but currently there is only a kick start with some
basic tests available. The good news is that I now learned enough
to do some sensible work alone in the coming weeks.
- I had some discussion with Frederic who did a really good job on
Debian Science stuff. We discussed several enhancements for the
webtool also how to include other meta packages nicely. The problem
of Debian Science is that for some fields simply meta packages of
Debian Med or Debian GIS are listed as dependencies which perfectly
works for meta packages but looks totally ugly for the WebSentinel.
So my Idea is to implement a "Meta-Depends-SVN" tag which points
to a tasks file of an other CDD parses the dependencies of this
tasks file and puts the resulting depends into the WebSentinel
page.
See usage hints in the Wiki at
http://wiki.debian.org/DebianEdu/Extremadura2008/WebSentinelHowto
3. Names and definitions
We had some very heated discussions about names. While all involved
people perfectly agreed that we should change the name of what we
call CDD now because people get some imagination about this name
that does just not to fit what we had in mind for the thing we are
developing.
- I tried to find some criteria for the three different things that
just exist. We should turn these in correctly worded definitions
that attach reasonable names to the things which is necessary
to enable a senseful discussion.
- We found some usable names but we did not attached them to a
thing as a real definition.
For a summary see the Wiki at
http://wiki.debian.org/CDDNamingProposals
4. Bug fixing and others
As usual some bugs were fixed in some of our Debian-Med packages
and some packages were taken over to group maintenance.
5. The mysterious knocking noise. ;-)
As partipiciants of the meeting know we had some strange knocking
noise in the meeting room. I was corious about this and started
some statistica about the frequence of these noises and the number
"booms". Some people tried to make fun of myself claiming that
I would be the only one who had heard this. But I can assure you
that my wife is perfectly living and I have clear evidence that
others heard it as well and we have even recordings done with OLPC.
The OLPC was especially helpful in proving that there are
actually two different noises happening. I would think that it
was due to my mental strength that when _I_ did the statistics
we had some regular pattern in all this stuff: There was one
knocking sequence with 7 booms in a row that happened exactly
at 4 modulo 10 minutes (so every 10 minutes when you had a four
as last digit). This pattern was perfectly regular over the
whole time of my recording. There was another noise that startet
at a certain point of the day when a 2 was the last digit, but
it was delayed over time. So the break between two occurences
was a little bit more than 10 minutes which leaded at about
22:00 to the effect that both noises intered each other which
was quite an experience! Most people missed this point because
they came back from dinner a little bit later. Then the other
noise (we called it the "wrong" one) happened after the right one.
I would not like to spend your time with more details except
mentioning the fact that on the next day (when my paper about
the phenomenon was ready and published) some other people took
over the work of writing down the noise-log. It turned out that
their mental skills was to week to create such a strong pattern
because neither the frequence of the events was stable nor the
number of "booms" was konstant as at the day before.
Well, for those who are crazy enough to find this interesting
have a look at
http://wiki.debian.org/DebianEdu/Extremadura2008/Knocking
but you might be interested in spending your spare time otherwise -
I had a long trip to spend my time on writing down this stuff. ;-))
See you at the next meeting and thanks again to the organisers
Andreas.
--
http://fam-tille.de
Reply to: