Mechanical

Using LiveJournal Code

For work, we're building a system which allows people to specify an object model in Perl code and have the database get automatically built for it. Yes, I know this has been done before, but we do some nifty tricks to push business logic into the database layer. Before those of you who know about the "object-relational impedance mismatch" choke, I have news for you: the impedance mismatch is a myth. Seriously, it's a complete myth. There's an object-dbms impedance mismatch, but that's a rant for another day. Suffice it to say that if one understands and properly applies relational theory, the apparent impedance mismatch is significantly lower than folks think.

Returning to LiveJournal, you may have noticed that it holds up very, very well to severe server loads. Many sites get taken down by Slashdot stories like the one about my identify theft, but LiveJournal held up quite well. Part of the reason for that is they created software called memcached. This is a nice tool which makes it easy to cache data and take the load off of the database. Unlike many open source projects, this one turned out to be phenomenally easy to compile and use. Two thumbs up!

Since we're also using Catalyst to map actions to commands for our Web UI, we're using Catalyst::Plugin::Session::Store::Memcached to manage our sessions. Once again, it turned out to be very easy to use and install. This makes Ovid a very happy programmer.
There's no object-database mismatch in some db-oriented ools (several on OS400 in particular), and you can obviate the mismatch with some pre-work in some dbmses (eg Postgres).
Say, good on yer! for writing stuff for Perl Hacks. I wrote some elisp to complement your run_tests_from_vim.pod but found the idea kind of clunky. I often have an editor that I launch once and use for everything. I get the impression you go to your module's directory and edit it from there. The paths you wrote to get to C<../lib> appear to be relative to the module and not from C<~> which is somewhere else entirely.
What did you find clunky, the elisp or running tests directly from the editor? I've been tempted to learn Emacs just for elisp.

As for the paths, I did that because I have a set of bindings that work anywhere, regardless of whose box I'm on, just by following a standard practice of where I edit from. That means I don't have to worry about whether I'm hopping on a client's box or someone else's.
I found it clunky that I'd have to start an editor from a particular place. Like I said, normally I start it from some menu in X and I suppose that means F<~> is my current directory.

That same editor stays open and handles all the myriad projects and work I do throughout the day so it'd be downright strange to be quiting and starting new editors just because I'm working on something different.

Oh well. This just highlights a difference in how we work.
props
Props to you for giving props to others. Glad your project is going well.

Fairie G
I've been intrigued by the idea of Catalyst - especially having done a little with Ruby on Rails and wishing I were in Perl. Any particular site you'd recommend for some good info?
Well, you can check out catalyst.perl.org, but Catalyst definitely has a "rough" feel to it. The documentation is spotty and the tests are even worse. So far I've asked, more than once, where tests for "X" were only to be told "there aren't any".