I'm way too attention-deprived to "live blog" at a conference, so here's some of my notes transcribed from paper to the interwebs…
Zed Shaw's opener
- Stop doing math demos as benchmarks – math's not the problem, IO is (talk to the database, write to a file, respond to client)
- Amdahl's Law: throwing more hardware at bad code doesn't really help – better code trumps faster hardware
Obie Fernandez' "Worst Rails Code"
- There's a Ruby anti-patterns book in the works. If it includes "how to do it better", this could be cool.
- Suggested: There should be more "how to re-factor this" blog posts, take up the Rails Way slack.
I didn't particularly enjoy Obie's talk. It seems like an easy "win" to look at someone else's code, say "FAIL", and laugh; it's harder to point to a better way.
The new and inexperienced write bad code, because they don't know better. But the way they learn is by writing; does publicly shaming the bad code make them learn faster? I don't think so.
Sam Aaron "Code Aesthetics"
Sam's talk was very conversational, I didn't take very interesting notes. Flog looks interesting.
Ninh Bui & Hongli Lai "Phusion Passenger (mod_rails)"
This was probably my favorite talk of the day; a good blend of practical information (including how-to videos) and high-level theory. Passenger 2.0 looks really cool.
Charles Nutter "JRuby"
I don't have a need of JRuby just now, but I was nonetheless impressed. I expected setup and deployment to be much more complicated than it looked like here. I noticed that Charles frequently used the expression "package it up & ship it out", whereas the mainstream Ruby community may be more accustomed to "clone the latest from git and check it out" — it's a different metaphor for software delivery.
- There seems to be a rift growing between the "write what you know / optimize programmer happiness / programmer time is expensive" camp and the "performance is an issue / 'worst rails code ever' / optimize now" camp. The last group seems to spend more time in extremistan than the former, for better or worse.
- A lot of time yesterday was spent talking about performance. Maybe this was an accident, but I think it's reflected in the larger community (there are how many projects working on newer/better/faster Ruby VMs?) Is this really a good use of our time?
- Here's my short list of "what we aren't talking about, because 'performance' has taken all the air out of the room" — testing, globalization, contributing to Rails, alternatives to ActiveRecord.
- Both Obie and David are bothered by re-inventing wheels. Haven't we been through this before? If joining a programming cult doesn't save me from inane dogma, what will?
Photos from danny_l