EuroPython:

Holger, can you introduce yourself briefly?

Holger: 

I am Holger Krekel, a free software hacker.  In another life, I am
living with a bunch of like 50 people, mostly artists, musicians and
computer people.  We bought a large house (http://trillke.net) together
with a nice park and are self organizing things.  It's certainly a
major project on its own.

EP: How long have you been programming and how did you come to be a
    "free software hacker"?

I started somewhere in the late eighties with assembler in a gaming
company.  I went through several jobs, studied computer science and
did a lot of consultancy for banks.

I arrived at Python in 2001 after having done a CORBA transaction
service the years before.  I got seriously annoyed at the lack of
innovation in those banks.

In the late nineties I more and more participated in the free software
culture.  From then on I learned a lot more than the years before.
The first EP was great to meet people and it got me involved a lot
more.  It basically led to founding the PyPy project with Christian
Tismer and Armin Rigo in early 2003.

EP: The PyPy project has attracted an impressive amount of interest
for such a young, speculative project.  Can you say a few words about
what it is and what it is trying to acheive?

The basic idea is to use a very-high-level language like Python to
implement a Python interpreter, and encapsulate decisions about memory
and threading models cleanly.  Additionally, Armin Rigo has been
inspiring us to head for the "impossible" goal of making PyPy faster
than C :-) Finally, we should be able to have a common code base from
which we can specialize/generate interpreters for specific
environments like embedded devices or high performance database
servers.

Python is just so much nicer to express ideas. Generating machine
code directly is surely interesting in many ways.  Not only for
performance but also for understanding what's going on and lowering
the barrier to experimenting with it.  Most importantly though, the
PyPy developers are having a lot of fun doing various kinds of stuff
related to PyPy :-)  Note, that it's also a research project so
we have to experiment quite a bit.

EP: The pypy-dev mailing list had something like 400 posts in its first
month, before any code had been written.  What is it about the project
that gets people's attention, do you think?

Many people think that Python is too slow and the JVM and its JIT
compilers beat the hell out of PVM (Python Virtual Machine).  I mostly
disagree but i think it's one of the driving forces for the interest.

Also every python coder likes coding in Python and so rewriting the
interpreter in Python sounds interesting (rightfully so!).  

And then i guess that the names of Christian (Stackless) and Armin
(Psyco) also drove in quite some attention (rightfully so!).  And
someone named Michael Hudson also participated early, which helped
getting attention, too :-)

After all, free software development is a lot about the *people* being
involved not just the technology.  In this respect, EP is a premier
venue for meeting people.

Let me add that i got my first python job offer at the first EP and am
still working witht he great people at Infrae in the Netherlands.

EP: For me, PyPy is also interesting in two less 'technical' areas:
sprints and the EU funding it has recently been pledged.  Do you have
anything to say about either of these developments?  Would you
recommend other projects go through the funding proposal process?

Well, we are in negotiations with the EU and it's a tiresome ride to
be honest.  It certainly took quite some time, especially in 2004. I
think it's a little early to speak of recommendations.

The EU's structures are not all fit for free/open software development,
especially when it comes to management and money issues.  But the EU is
definitely trying to allow open source projects to receive funding.  In
the PyPy case, we are applying as a research project but there are lots
of other programs you might apply to.  It's tough to see through all the
rules and maintain the spirit and ideas of the original project.  I
think we managed quite ok but only in the next months will we know if we
did it right.  Then I will be ready to recommend something. 

EP: So it's a case of the EU and the free/open source community both
needing to learn about and adapt towards the other?
EP: and you didn't say anything about sprints :-)

A word about sprints: they are great as you can learn a lot from each
other.  And you get some real coding done and see how others do
it. This has been the single most different thing on how i used to do
development before.

But there are problems with sprints that nobody much talks about.
Like getting everyone to the point where he feels comfortable and
enjoys participating.  That doesn't come for free and requires some
thought on the part of the organizers and participants.

Also, getting along on the personal level is really important.
Sometimes I think we should adopt techniques from "social groups" who
play games to get to know each other.  Technical people are often
pretty hesitant and shy to play such socializing games.  If nothing
else spending the nights together helps :-)

Back to EU vs. open source communities:

The open source community has the advantage that it is used to
understanding and working with and around "foreign rules".  The EU has
a large set of such rules.  The challenge is to understand them and
make them fit to your project.  The EU should learn to work with
'flatter' hierarchies, I think.  They think in terms of "one
coordinator", "one manager" and so on in the classical sense.

Of course, you have to see that lots of people are trying to exploit
EU funding.  It's a huge industry so it's understandable they try to
control and regulate who gets money for what.

Also, it's not clear what the "EU" actually is apart from a large
organizational body.  That's why it's far from easy giving them
feedback or suggesting improvements.

EP: OK, let's step back to why we are attending EuroPython.
EP: What do you think is Python's single biggest selling point?

It's both extremely productive and easy to use compared to c++, java,
perl, etc. It's the combination of 'productive' and 'easy'
that makes it, I guess.

Though everyone points at the "batteries included", I am not that sure
that python's library is still state of the art but that's a different
topic.

EP: And the obvious next question, what do you think is Python's
single biggest weakness?

It doesn't leverage and support the distribution of useful modules.
There are some interesting efforts but it's still too loose I am
afraid.

The barrier to collaborate code-wise should be lowered as much as
possible.

EP: Is this something that any community gets right, though?  I guess
the perl and Java commnities might be places to look for inspiration,
but I know little about them.

Obviously, there is CPAN but I don't know about Java.  I don't think
that the free software spirit is too strong with Java.  Putting away
comparisons: There are no deep divisions in the python community and
that's a very good thing.

But if you want to use 3 modules from other developers while offering
your own: what do you do? The choices are all kind of limiting and
inflexible unless we find a way to have a sane "package and dependency
system" that helps organizing this.  And i happen to think that
subversion and python should enable us to do it nicely. Today, we
have better technologies for distribution and versioning than ten
years ago and we should definitely leverage them.

It's a tough problem for sure, and i guess no comunity has got it
right completly.

EP: Do you have any other Python fans hiding away at Trillke?  Are any
of your housemates coming to EuroPython in June?

Unfortunately not really. Some of them are doing some python scripting
and enjoy it.  

Around 5-10 developer friends kind of got into python but they don't
feel like going to conferences.  Some of them said it's because of
lack of business opportunies, but i think they mostly think that it
would be "over their head".  Maybe they are just too lazy (that must
be it I guess :-).

EP: And finally, what do you expect from EuroPython this year?

Well, i am going to be two full weeks of hacking and conferencing in
Sweden. I expect lots of fun and interesting people and projects to
meet!

And i also hope to get feedback on some of my non-PyPy projects.
Especially attracting new developers for the 'std' library that i am
initiating with Armin.

EP: Thank you for your time.

Thanks have to go to you and the other organizers!