Wow!
YAPC::NA 2006 was without a doubt the best conference I have ever attended, both due to the quality of the talks (many of which I missed and will have to watch the DVD to see) and the quality of conversation.
Because I'm currently spread fairly thin at the moment (between JSAN, PITA, Module::Install, CPAN, Perl 5 core things, TAP rewriting, TAP::Harness::Straps issues, Perl 6 core things, TPF and grant conversations, Vanilla, Strawberry and other win32 Perls, a dozen other modules people want to talk about, plus general chatter with various people) I've had no time at all for journal updates.
My (it seemed to me) incredibly well attended "Nothing can possibly go wrong" talk went extremely well, and I had wave after wave of positive comments from people. I must have got a really lucky slot too, because by the end the number of extra people coming in during the talk meant it was standing room only at the back.
The general topic area (how to screw up when designing big software systems) certainly seems like a very rich vein to tap for future talks. I barely scratched the surface on this first round of "Problems".
So stay tuned for a "Nothing can possibly go wrong... again" talk at next year's YAPC::NA.
The other "PITA - Rediculously Large Scale Testing" talk unfortunately lined up against Andy Lester's "code debt" talk and CLKAO's svk talk, and so wasn't as well attended, but the talk itself went over really well (scatalogical movies included) and there seemed to be a lot of excitement around the direction PITA is going.
Now that the hackathon is tailing off and I have time to actually write a journal entry I thought I might as well summarise the things that came out of this for me.
------------------------
After "Nothing can possibly can go wrong" a number of CPAN authors have mentioned they are going to start opening up their repositories to other authors. This is very encouraging, and I'd love to see repository access (at least to other CPAN authors) start to become the rule rather than the exception, especially for anyone who has 10+ modules and is running into maintenance scaling issues.
------------------------
On a similar topic, I've now added write access to my repository for another 5 authors. Most notably, both Jeff and Chris from Perl::Critic have got accounts so they can fix any PPI bugs that are causing them problems, rather than just filing rt.cpan.org bugs.
------------------------
YAML::Tiny should start moving forward again. David Romano has write access and is interested in helping, and Ingy has also gotten interested (although his interest seems to be in writing one giant regex that can parse YAML) :)
So hopefully in the fullness of time we should have a decent light-weight parser for things like META.yml files, Plagger/et-al config files and other similar trivial YAML uses.
------------------------
Perl is looking like it has a hiring problem. Prices are high at the moment, but the companies like TicketMaster that want to hire dozens of coders simply can't find the numbers locally, with other good Perl people are spread thinly across the US and unable or unwilling to move.
This matches up against rumors from Australia and Europe, although in Australia the lack is of top level people, rather than filling the low ranks.
At the same time, there was a good amount of talk about how companies might participate more, and also benefit more. But I don't want to preempt anything here.
------------------------
Dave Golden is officially insane, and seems to share Audrey Tang's ability to just dive into a problem on 5 seconds notice and have finished it before you've finished having a conversation about it... during a talk.
Compared to me, he's put in an embarrasingly large amount of effort on getting Vanilla build 4 and Strawberry alpha 1 and ready, and I've mostly been limited to design discussions and general strategy due to aformentioned thin-spreadedness.
You should look out for Perl::Smith soon, which we should be able to use to build just about any custom Perl distribution we need to (although it might not work cleanly until 5.8.9 is released and syncs up to some of our changes discovered in Vanilla 1-3.
------------------------
In similar news, Win32 Perl (or rather win32.perl.org now hasa logo. It's been passed onto the Perl foundation lawyers via Allison Randal to verify that the use of the four windows colours in the circle is ok and won't violate US trademark law.
Following Randal Schwartz's advice, we're also going to be taking up some more TPF laywer time to examine he way we are building Vanilla and Strawberry (et al) and make sure that our combination of the various components with different licenses is done in a compatible way, or if we need to make some changes to acommodate what we are doing.
Apparently there's been problems with other projects in the past, but since our goals are to never patch anything and always use the defaults, I'm hopeful there won't be much we have to do.
------------------------
Jonathan Steinert is also insane :) He's also a huge bundle of excitable creative energy, and is going to be helping out to finish the all-important PITA::POE::SupportServer, which is currently blocking the primary coding path for PITA.
------------------------
Pete Krawczyk has completed his image auto-install script, and is in the process of putting together a "gold master" reference Qemu image, for use both in testing, and for doing very basic experimental testing runs.
He's also registered pitatesting.org, pitatesting.net and pitatesting.com. The .com is there just in case, but we're expecting to use pitatesting.org as the development website and resources (mailing lists etc), and pitatesting.net for operating infrastructure, such as image distribution, large databases, bittorrent servers and so on as they are needed.
I'll probably be hosting it on the DreamHost box we currently have win32.perl.org on, but once we start building the reference image library that might need to change.
------------------------
After catching up with Casey West, things are going to start moving on JSAN again. I've moved in and rereleased my four JSAN distributions, and it looks like the new shell will be tentatively blessed.
After finally having the mirror.conf patch accepted, we also now have auto-validating mirror URI's via JSAN::URI. This should bubble outwards soon to the shell, especially now things like File::UserConfig are available.
-----------------------
I kind of understand Moose now... and I think I get Meta-Object Protocols, although I still completely do not understand why it's called a "Protocol" (well, unless you turn your head sideways and squint).
In addition, the Moose guys are undertaking one of the larger underground marketing campaign ever for Perl 5 modules, possibly to offset the fact it's called "Moose" instead of something more corporate friendly :)
I don't buy a word of it, even if I did agree to take some "moosey" photos.
-----------------------
In more general (and happy) news, there seems to be very strong corporate interest in PITA, and a couple of people even said they got permission to go to YAPC solely on the basis of the PITA talk. So not only do things look good for PITA in general, but they look pretty good as the basis for a supported product.
Apart from Pete, I have volunteers to do trialing, and some people interested in putting together images, and a couple of potential driver writers.
-----------------------
Believe it or not, that's only half of the wrapup, but Andy beckons me away to some Pizza place that is meant to be the pinacle or "Chicago Deep Dish Pizza".
I still completely do not understand why it’s called a “Protocol” (well, unless you turn your head sideways and squint).
I bet that’s a Smalltalkism. Since classes in Smalltalk tend to have so many methods, they are grouped into protocols, which are collections of methods of a single class, associated with a keyword phrase. Some protocol names are common and have agreed-upon meanings, such as number, observing, sequence or private.
But I can’t bet too much on it, because the first MOP was part of the Common LISP Object System, not Smalltalk. (Smalltalk-80 was the first version to sport a MOP.)
