Wednesday morning I saw the talk on Mail::Box. Interesting. One of the things I've decided I like are modules that just do the right thing in most situations. No massive interface. No jumping though hoops. One of the examples of Mail::Box was the coercion of Mail::Internet messages and it's ilk into the right object type. Lots of lazy work going on behind the scenes (for example, using 'partial objects' that can upgrade themselves if they need to do more work.) You could force the user to do this and worry about it...but the module handles it all behind the scene. Nice.
james' Pixie talk this morning mention that Pixie does this kind of thing too - it pulls objects out of the ODBMS and stubs them with proxy objects. I don't like having to think about things. I sometimes get them wrong. Modules that do the right thing are a Good Thing...It means someone else did the thinking for me and codified it into code.
The other thing I think I like really well are very intelligent questions at the end of the talk that are answered really well without excessive detail. Not questions about the typos. Not questions about people that haven't understood what's been said. Questions about big problems. For example, james didn't actually cover the stubbing process - that came out of a question. I really like speakers that say "So the question was..." and rephrase the question so it's clear how they're answering. I don't like it when speakers try and explain a complex thing. I like it when they say "It's complex and you could do something with foo...infact a quick example of that was when we did..." Describing code is a hard thing that should not be attempted...especially when both of your native languages aren't the one you're communicating in.
Schwern did a lot of good question answering yesterday afternoon in his Testing talk, and the whole thing was very entertaining. I was a little disappointed with what was actually covered though. To be honest, the main problem was that most people hadn't seen (or read) his Test::Tutorial talk that was essentially required reading. So he had to go over that a lot and the whole thing became disorganised. Greg did do a wonderful impromptu presentation half way though the talk though - excellent considering schwern had only told him at the start that he expected him to present with slides.
In the evening I headed off to the speakers dinner that was much fun. The place we went to was so much better than the pub we went to on Tuesday night. The food was good, the service was excellent, and I could be convinced that this whole table service is a good idea after all. I was introduced to the system of the staff writing things down on beer mats which you pay at the end of the evening, which seems a lot better than the previous system of the server trying to remember everything in their head. I like people coming round and asking you if you want a replacement beer when your glass is empty. Well done for Munich.pm for organising such a great night.
The trouble with staying in the hotel with the other London Perl Mongers is that when you get back you can find them all drinking the bar downstairs. The other problem is that Greg will try and convince you that everyone drinking brandy in his room is a good idea. Do you know how hard it is to buy painkillers in this country?
This morning I saw cwest's Template Toolkit talk. It was very good and I'm glad that he did it. Now when I do my talk on Friday that says "I'm not going to explain TT syntax" and anyone complains I'll be able to say "well you should have gone and saw Casey's talk then." Eeek. Talking. Should find some time to tweak the slides somewhen.