Slash Boxes
NOTE: use Perl; is on undef hiatus. You can read content, but you can't post it. More info will be forthcoming forthcomingly.

All the Perl that's Practical to Extract and Report

use Perl Log In

Log In

[ Create a new account ]

Alias (5735)

  (email not shown publicly)

Journal of Alias (5735)

Wednesday April 19, 2006
10:52 PM

I'm just so sick of the web...

[ #29383 ]

So it appears that apart from the dropped PPI talk, both my other YAPC talks have been accepted.

While the PITA talk should be a pretty normal affair, I'm hoping that the second one, Nothing Can Possibly Go Wrong, will be a little different.

If it works, it will end up something of a cross between a really theatrical rant about some of the biggest high-level, non-techical, but non-obvious, screw ups in some of the most popular CPAN modules, and a mea culpa on the places where _I_ screwed up big time and didn't see it coming.

For example, 2 years ago, I "solved" the web.

I thought I had it all figured out. The best ORM layer yet in Perl (still), by far the best widget system (although Ovid is heading in that direction now too), deep, highly DWIM integration with Template Toolkit, properly configurable, auto-generating, auto-upgrading, trivial to churn out multiple web front-ends to a complex model, in a single day. Magic so that even designers who don't know HTML can work with the templates. Magic so you don't need the help of random hosting company's sysadmins, ever. Magic to ease maintenance, magic to catch your deployment screwups.

Dear god, I even got it writing it's own unit test suite, 10,000 cases per application over 4,000 to 5,000 lines of code on average. Once Isotope comes out and I can rip out my pretty average engine and integrate that in, it's going to be pretty close to ideal in my opinion.

It may sound egotistical to say that, but then I'd gladly put it into a getting-the-job-done competition against all comers.

It's ruined me for the web, I just can't get excited about it any more. Maypole, Catalyst, Rails, Jifty, whatever-comes-next, all Yet Another Web Thing. Still all with problems I'm sure they stil don't see... (although in Jifty's case it's nice to see someone else discover form-based dispatch finally, well done)

But with a few exceptions, you've never seen it, and you most likely _will_ never see it, because you can't see it.

Because in retrospect, I screwed up big time, high-level, non-technical, and non-obvious until it was way too late to fix it. (and no, not a legal thing either).

I solved the web, but I did it in a way that means I can never release it :(

So you'll never see it, just some of the most minor of minor spin-off parts, like Class::Autouse and Config::Tiny and Class::Inspector, and two dozen other bits and pieces.

And so hopefully showing you MY 50,000 line of code screw up (and a few other smaller screw ups) will make the talk less like a rant, and more entertaining, in a train wreck kind of way.

So if you aren't coming already, I invite you all to YAPC::NA, to witness my train wreck, and hopefully to learn a few things so you don't crash your train.

It'll be so horrifying, you won't be able to look away! :)

The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
More | Login | Reply
Loading... please wait.
  • Hi,

    make sure your talk is recorded and made availlable online for all of us who can't come.

    • My company is planning to do the presentation capture and production for YAPC::NA this year. Here's an early sample of what the content might look like: []
      • That looks really interesting.

        How does it handle being overloaded?

        For example, my current generation of talks (one big one, and three or four lightning talks) average out at around a "slide" every 6 seconds. So that's something like 220 slides for the 50 minute talk.

        And after realising that the talks are getting closer and closer to being a video, this year I'm going to try something unusual and include ACTUAL video. The big PITA talk, at 80 minutes long, is expected to have 15-20 minutes of TV-quality vide
        • 220 slides is fine. We can easily support a slide every 2 seconds with the current technology, and we've got further enhancements in the works.

          As for video, we're working toward that. One of our present solutions is to use a subclass of Net::VNC to capture local screen video. Those can then be converted to QuickTime movies or a series of image files for the resulting presentation. We're currently capturing at about 5 fps for typical desktop app motion, but there are certainly improvements to be made.

  • If you're talking about the product I think you're talking about, it's a damned shame you can't release it. I was very impressed with what you showed me.