Steve Mynott

To me It seemed a particularly good FOSDEM for both for Perl5/6 and
other talks although very crowded as usual and I didn't see the usual
*BSD or Tor stalls. I was stuck by the statistic that there were
about 500 speakers from many thousands of people so of the order of
one speaker per tens of attendees which is very high.

Videos are already starting to appear at

http://video.fosdem.org/2016/

On Saturday I started with Poettering and systemd which was a keynote
and perhaps a little disappointing since he usually is a better
speaker and the audio was a little indistinct. systemd had won being
used by all distros except gentoo and slackware. They were now working
on a dns resolver component which supported DNSSEC although in
practice validating signed zone files would slow down browsing and
currently only 2% of websites had it activated. He didn't mention
strong criticisms of its security by crypto experts such as DJB.

The most amusing talk was Stark's retro running of Postgres on
NetBSD/VAX which exposed some obscure OS bugs and was livened up by a
man in an impressive Postgres Elephant costume appearing. We later
spoke to Mr Elephant who said he was both blind and very hot at the
time. I then went to the Microkernel room to hear about GNU/Hurd
progress from Thibault since this room is usually "OPEN" and he's an
excellent speaker. I noticed even this obscure room was quite crowded
as compared with previous years so I'd guess total attendees this year
were high. He stressed the advantages of running device drivers in
userspace as allowing more user "freedom" to mount fs etc. without
root and improving kernel stability since the drivers could crash and
restart without bringing down the kernel. In previous years he had
talked of his DDE patches allowing linux 2.6 hardware drivers on Hurd
and this year he was using the NetBSD Rump kernel under Hurd to add
sound support with USB support promised. His demo was RMS singing his
song on his Hurd laptop. The irony was he needed to use BSD code on a
GNU/BSD/Hurd system to do it! There had been some work on X86-64 Hurd
but it wasn't there yet since he needed more help from the community.
I then saw some lightening talks (actually 20 mins long) including a
good one on C refactoring.

The Perl dinner on Saturday night featured the usual good food and
conversation and the devroom was on Sunday. Ovid spoke about Perl 6
and its advantages (such as being able to perform maths on floats
correctly). I had a python guy sitting next to me who admitted he had
never been to a Perl talk before so that was a success in reaching
someone new. Will Braswell spoke next about his "Rperl" compiler
which translated his own quite restricted subset (no regexps yet and
no $_) of Perl 5 line by line into C++ in order to run some of the
language shootups benchmarks (a graphical animation of planetary
motion) at increased speed. I'd not seen Will before and he was an
excellent speaker who left me more impressed than I'd expected and I
hope he gets to YAPC::EU in the summer. I saw some non-Perl stuff
next for variety including a good one on the Go debugger Delve which
was aware of the go concurrency and could be used as a basic REPL. I
returned to Perl to see Bart explain some surprisingly simple X86-64
assembly language to do addition and ROT13 which he interfaced with
Perl 6 using NativeCall (although it stuck me that the
CPAN P5NCI module on Perl 5 would have also worked).
Again an excellent talk and a good start to the a
run of some of the best Perl talks I'd ever seen. Stevan Little's talk
was one of the his most amusing ever and perl wasn't really dead.
Sawyer also did an excellent promotion of Perl 5 targeted at the
people who maybe hadn't used it since the early 2000s explaining what
had changed. Liz finished with her autobiographical account of Perl
development and some nice short Perl 6 examples. We all ate again in
the evening together my only regrets being I'd missed the odd talk or
two (which I should be able to watch on video).