Glad to be home.
First this morning, I went to the MVC talk by Perrin Harkins. He compared three web-app frameworks: CGI::Application, Catalyst, and Open Interact 2; his slides may cover other MVCs he didn't have time to discuss too. This talk felt like it should've been longer than 40 minutes, given the content. He was able to illustrate a bit of sample code, but wasn't really able to compare and contrast as deeply as he has in his online writings. I'm left wanting to learn more, but not knowing if I actually have any more additional knowlege than the little bit I wrote below.
- Catalyst features include very flexible URL handling, which you won't get in OI2 (without hacking). There are tools to generate lots of stub methods, for easy startup. However, documentation on this is shaky; perrin had to read sourcecode for 3-4 modules to get that to work. There are good plugins available, which will illustrate ideomatic code for you. Data relationships are stored in perl modules, such as via Class::DBI::Loader::Relationships perhaps.
- CGI::Application is the most skeletal framework of the three; you end up rolling things yourself or at least, writing enough to piece together some of the (many) good plugins available.
- Open Interact is quite automatic. You turn it on (it runs its own webserver at the outset), click a few buttons, and you have a MVC. There is more abstraction, for those plusses and minuses (he described one level which he found quite irritating; I didn't catch what it was called...) OI is quite appropriate for large-scale projects, where it will be worth the effort to learn. OI uses lots of configuration files (which I believe is where data relationships can be defined).
So, I have some information to go on for my next web app project. It's frustrating; the last MVC I used was Maypole, which I never quite got working well enough to do anything beyond the examples; since then I've re-invented wheels a few times on small db-backed projects, and I'm loathe to do the same again, but at the same time it's not completely clear that the existing frameworks will satisfy either. Guess I need to dive in and try them...
I skipped the next time-slot to check out and chat with people a bit; I was unhappy about missing Casey West's talk on development tools, but luckily (?) he got bumped due to over-runs and got to give a shorter version to everybody in the afternoon. Which was an awesome talk. He ended up with nine wonderful development tools I am pretty excited to learn about:
- SQL::Translator - goes from any DBD to any other; and will even make images, too.
- HTTP::Server::Simple::Static - lets you run a no-muss, no-fuss HTTP server in your web directory, for quick testing.
- Devel::Cover - How much code are your tests covering? Not only provides answers, but spits out very pretty HTML tables of modules and functions which you can drill down into.
- podwebserver - another standalone POD database, but looks quite fully-featured. Will show you the source, and tidy/colourize it for you, too.
- perltidy - speaking of which, perltidy will allow your coding team write using each person's own preferred style, then you can run it through the (highly customizable) perltidy script to make it all standardized.
- Test::Pod::Coverage - self-evident?
- DBI::Shell - command-line tool, db equivalent of perl -e (or possibly perl -de1), with handy switches.
- Module::Refresh - nit sure what this one did
- CPAN::Mini - 600MiB of current CPAN goodness, as opposed to however many GiB are actually available.
After the morning break, I watched Ingy discurse about Data Driven Testing (so, writing tests in "0 lines of code" which actually means writing config files in a very easy to understand format) and Perldoc, his ideas for making POD suck less. I said "discurse" because he has an amazing ability to go off on tangents that are both trival and very important, and if you're not paying at least as much attention as he is, you might miss what he actually said. For example, I completely failed to take notes during them because they were so entertaining. On the other hand, he wrote the URLs for us to follow along; but I missed it. (Edit: and Autrejus told me what it is, below.)
Lunch was with other KW Mongers, down at the Bon Marche on Bay Street. We debriefed a bit about the morning, but I'm not confident enough about what people said to repeat it here. (See what I did when I echoed Allison on Monday?)
What else. Oh, lightning talks.
Of particular note for me to look up [this is not meant as a slight against those I don't mention!]:
- The list of all 16 talks is here.
- Regexp::Compost does automatic Regexp decomposition to make them predict /future/ matches as well. Sounds crazy but the demo seemed cool enough that it might just work for some applications (eg., monitoring web logs). By firstname.lastname@example.org.
- SPF was invented at YAPC::NA Florida (03)? I had no idea. Anyway, Meng Wong, the creator, is working on two new projects, Domain Keys and "The Karma Project" for improving site-to-site trust, and "RSS email" as a complete reversal of the ideas behind SMTP. URL: somewhere here and possibly here.
- Not Just Another Perl Hacker: The Perl Foundation is soliciting nominations for the best technical innovations in perl. talk to dha.
- wikis and cities: cityguides.com
- there was entertaining juggling. As opposed to lame juggling, which would've happened in other fora, but never at YAPC.
Must go eat dinner now; more later (for some value of later) when I have a chance to look for stuff I only hinted at above... Comments welcome of course; it might be a few weeks before I respond though, because I'm off to the wilds of Virginia till the 12th. [edit: fixed URLs, 6/30]
Thanks to all the people who made YAPC possible this year (from the organizers to the speakers to...) It was a lot of fun to help out behind the scenes, such as I was able to. Looking forward to next year.