Debconf - trip report
[ As a Canonical staff member I get some conference leave, and I chose
to spend mine at Debconf. Sensibly, we're asked to write a report
of such a trip. I thought debian-project might be interested to
read it - it basically says more or less the same things I'd say to
a DD about DC7 - so I've cut-and-pasted below. NB that of course I
won't report here in public on anything that happened at Debconf
that was confidential either because of my employment or just
because of who I spoke to and what it was about. ]
Last week I was at Debconf 7 in Edinburgh. Here's the main points I
remember from the trip. If someone reminds me of a conversation or
BOF I've forgotten I may post another paragraph or three ...
Spread over the week I had conversations about multiarch, with various
people including Steve Langasek, James Troup (in his capacity as
Debian ftpmaster) and others. Multiarch is a family of proposals for
dealing better with the amd64/i386 split and perhaps also with various
cross-execution setups. Those who've heard me on the subject know
that at least some of these plans don't have, shall we say, my full
support. But from discussions at Debconf I think there may be a
proposal floating around to do something much more sane.
I'm hopeful that something not wholly insane can be done to sort out
this situation which I definitely agree is currently a mess. Steve
promised me that nothing would happen that hadn't had a proper design
Archive testing of various kinds
There were a couple of interesting talks on archive rebuilding and
automatic testing. Debian seems to be accreting various ad-hoc
autotest setups, which are feeding back into Debian as QA
subprojects. This all looks quite good.
I mentioned autopkgtest there and got quite a bit of interest. If I
can get this running on the Debian archive then I think there's
definite interest from Debian maintainers in adding autopkgtest-style
functional tests to their packages.
Other dpkg stuff
We had some conversations about dpkg's revision control, and its lost
history. The series of different maintainers each did their own thing
and as a result the historical information can be hard to find. In
particular, if anyone knows where a copy of dpkg's arch (aka tla)
history can be found then I or debian-dpkg would like to know.
I reviewed an implementation of `filters' (excluding certain files
from being installed based on filename patterns). There are lots of
uses for this feature; in the particular case, embedded developers
would like to exclude documentation and other bulky materials to
permit installation on small systems. The author of the
implementation I reviewed will remain nameless; I hope that at least
they were able to take my comments as a constructive learning
experience. However I'm pleased to report that a different author
wrote a much better version which I'm expecting to see appear in
Debian unstable soon.
A couple of people spoke to me about dpkg triggers and seemed keen to
get an implementation asap. Sadly I had to say that with this
project's current priority within Canonical it's not clear that the
implementation will be complete in time for the current Ubuntu
Debian has been having some trouble recently dealing with socially
disruptive behaviour. The most example is the row which culminated in
the expulsion of Sven Luther.
There seems to be a rough consensus now to try to improve the
situation probably by setting up something like a Social Committee. I
had a number of corridor conversations with interested people and we
did eventually managed to get our BOF rescheduled so that most (but
sadly not all) of the people could make it - at 0945 on Saturday
morning. The BOF went well and seemed to produce answers to the big
questions that needed deciding; also, we had an attentive DPL who
seemed to be happy to rubberstamp what we come up with. Discussion
over details continues on debian-project.
Somewhat amusing was the juxtaposition of the two trademark BOFs.
The first was basically `what do we do about being screwed over by
people with trademarks' BOF, which was led by Eric Dorland and had as
its example case the problem with Mozilla which led Eric to rename
Debian's Firefox to Iceweasel.
Eric seemed to come along wondering if he'd done the right thing but
it was clear that the BOF (which was quite a self-selected set, of
course) was right behind him on that case and generally seemed perhaps
even more vigorous in its views than Eric himself.
General principles were discussed, including the fact that Debian
generally tends to ignore trademark problems until they turn out to be
real and bite us, and we will then consider renaming things; the BOF
seemed to think this was a reasonable approach.
Sun's new OpenJava(tm)(r)(?) was also discussed briefly here, and I
had some corridor conversations with one of the Sun guys; I don't
think there's going to be a problem but it's always wise to keep an
eye open and make sure everyone's communicating.
This was immediately followed, in the same room, by the `who and how
shall we screw over with our trademarks' :-). That was a much less
interesting BOF and concentrated mainly on some administriva. The
upshot as I recall it was that only really the word `Debian' (and the
`Official Use' logo - the one with the bottle, that people rarely use)
will be subject to any serious enforcement. The open use logo (the
swirl) will be MIT-copyright-licensed and the trademark used only to
make sure that Debian (and its downstreams) aren't themselves
prevented from using that logo.
On Wednesday 153-ish of us departed Edinburgh for a day trip to the
Isle of Bute. For reasons to do with railway ticketing (!) we had to
organise ourselves into groups of 4. We had a good trip there on the
train and ferry - packing into a normally-empty service out of
Edinburgh whose nominal seating capacity was 150.
Bute was very pretty. The group I was in went a-wandering over a hill
and through some footpaths, taking a very scenic and empty route to
Loch Fad. We were the first to get there and enjoyed the scenery and
quiet. A little after that we noticed the arrival of a horde of
Debconf attendees at the other end of the dam - and they had bought a
huge pile of provisions and set about a picnic which they kindly
invited us to share.
Mako gave a fantastic feelgood talk about how Debian is really
interesting to all sorts of people from outside the direct field of
computing, like sociologists, lawyers, voting reform advocates, etc.
It made us all proud to be part of Debian, and of course gave us an
insight into how what we do affects the world at large.
On a less good note, Edinburgh is very bad for bicycles. The cobbles
and rain together broke my laptop (battered too much in my pannier I
think), my back light, my front light, my emergency front light's
bracket attachment, my GPS attachment, my lock and some other things
I've probably forgotten. The laptop is currently with IBM being
fixed. So if you go to Edinburgh with your bike, make sure to
tighten all your screws and take enough screwdrivers etc. The
Debconf network admin team were very helpful in letting me borrow a
couple of things to make my bike useable for the commute to my
The night venue was a very nearby old church hall; I'm told it was
used during a recent intergovermental meeting as the organisation
point for the local anarchists and indymedia types. Debconf had a
jolly good time there, particular on the evening of the Cheese and
Wine BOF. Lots was drunk and lots of nonsense was talked. There was
even an organ, which our intrepid engineers fixed so that our
musicians could play it! Fantastic.