YAPC: The Sessions

Sat Jul 2 2005 09:52 MDT #

DAY 1

This con was huge for a YAPC - 400 people not this big since 1999. Maybe 2x last year's size.

Larry Wall, opened with a speech on building communities, which was (intentionally) long on questions and short on answers. I read it as a comment on the Perl6 / Perl 5 interaction and how the community was taking it. Though from my perspective that has been getting better with every patch pugs gets. Regardless a good speech from someone who has though long and hard on this.

This was followed with Alison Randal, of TPF fame who gave her State of the Carrot speech, mainly pugs, the new onion logo, and her signature poetry. Also mentioned was the 8 summer of code grants.

After that I did a talk by Peter Scott, of The Perl Medic fame. Lots of modules for testing and refactoring, but also some controversial statements along the lines of "Rewrite - it's your time", "It runs - BIG DEAL". Basically focus on the maintenance times - if rewriting saves you time do it. Of course all of the refactoring, or even maintenance follow the creation of a comprehensive test suite in his view, so this is less risky advice than it sounds. He also recommended a $SIG{__WARN__} that mails warnings to you for production code. And he suggests that use fatal is a useful debugging tool.

Next up was Ingy's speech on Kwiki. It was more a tour of his brain than a speech, but as it turns out that is what I needed. It was truly helpful, and I think Kwiki plugins, and modifications will be much much more useful. And beside you can read the slides, to actually learn about kwiki.

Next up I went to Geoffery Young's speech on the finally production worthy mod_perl 2. Very good and I'll seriously consider it on future projects.

Day 2

Autrijus on pugs, his perl 6 interpreter. An interesting speech - every one attended, so when he made digs about design - you could watch Larry who mainly laughed, and heckled a little. Basically pugs is Autrijus's toy. He makes the calls - and that means that is full featured, and fast to develop on. But it's not helping out parrot too much (though moving that way), and it's not fast. But it ain't bad and on some benchmarks it beats perl5. But unspec'ed features like Perl5 interop got in before macros and other hard stuff. Still useable today - and though he stepped back from Alisons 90% of it works number -- she was right in the 90% of what you want to use works sense. It's useful today for throw away scripts - and a good bit faster to write than even perl 5 for some of tasks.

Also the community is huge 5 months since the start, 100 commiters, 5000 commits, and something that works. Wowzah

P6 the language was next up this can't be summarize so read the p6 bible, this was followed by a some stuff on the parrot vm. Basically it's fast. It handles a lot and optimized for dynamic languages. It's starting to get stable. TCL, PGE, and pugs all target it. And speaking of PGE (perl6 grammar engine), it works. Missing some features and a little slow, but if you are looking for a heavy weight, unicode aware parser, that does all the tricks this is a good bet. It can switch out parsing strategies mid rule, and go from recursive-decent to shift-reduce, just like that. Bunches of tricks, and very nice.

Day 3

Started with Dan Sugalski's speech on memory leaks. Basically 5.7 started a massive leak hunt, and though that's wrapped up people are still hunting. Upgrade if you have a problem. But also IO::Handles are huge but rarely used. A Strings are almost always the problem. Big strings, in long running processes - since perl almost never returns memory to the OS. Slurped Files in list context cause many problems. Arrays alloc to the ending index. Lots of other trivia.

Then there where lots of little speeches, lightning talks till the keynote. The highlight where mainly module recommendations, though also of note:

  • Extending the debugger could be very cool.
  • Yuvals speech said basically if you want it robust, build a framework, and keep the biz logic code small.
  • Portable programming is dang hard - even if perl is a great choice to do it in. VMS, swahili, user permissions models, gads ....

The Keynote was not much of a keynote, but it did get slashdotted. Then a brief Q&A and it was done.

On the whole an excellent technical conference, amazing people, amazing speakers. It was also the best organized YAPC I've been to. On the whole very, very impressive.




Day 4 -- Post YAPC

I had to 3 hours sleep, woke up with a cold and composed the rather surly entry that was posted yesterday. Ooops.

Add your comments