Enjoyable and interesting talk. And although it was mostly about the problems with the SPL, I still got the overall message 'it's good, you should use it, but here's what's going to trip you up'.
Really interesting talk. It was nice to see the evolution of approaches taken, and some neat, and new to me ideas (e.g. checking encoding of the letter 'e').
I suppose there's not so much that's immediately actionable, as a lot of the tools used are internal to FB.
Ditto other comments - I think Glen mentioned that he'd been drafted in at short notice.
Good talk, a good bit to think about, some to disagree a bit with (or at least not do in the same way).
The bit about testing SQL near the end was interesting; the example showing hard-coded SQL seems like a bad practice. Perhaps the example could/should show either just a comment // do something SQLish here... // or suggest a more OO approach to SQL building.
Related to that, the example showing the test of the SQL demonstrated that the two approaches (SQL vs code) are consistent, but not correct (i.e. they may both return the same, wrong result). To be a little more complete a test comparing the returned result to a known good result would be a minor addition.
Coped well with unfortunate screen issues, I thought, and the content was very good, pitched at the right level for me, and well delivered.
One comment on the room, not the talk, for next year (if it's here again). It's frequently really hard to see the bottom half of the screen, so slides can be tricky to follow. I'd suggest speakers next year in that room try and restrict the main content to the top half of each slide - I realise that'll make it tricky, but I'd prefer flicking back and forwards between slides and be able to read them.
Interesting and contained useful content, but could have focused more on shutting down future hacks i.e. with htaccess & apache configuration settings.
Good introduction to some useful tools, seemed quite rushed but then finished pretty early
A good introduction to statsd, something I'd not used before. Well presented by an engaging speaker.
@ Matt Parker
The unit tests for the specification prove correctness. The query comparison proves consistency. The combination covers everything.