I’ve spent the last 6 years writing about performance and making Rails and Sprockets faster. If you can do better, give me a PR and not a put down on twitter.
Sam Saffron is an amazing person and contributes to the Ruby performance ecosystem in spades. I’m happy for his critique and his presence in the community. Jeff has done nothing in this space and by this tweet appears to know even less.
What exactly have I done? There was that one time when I decreased object allocation by 29% and gained 12% faster page rendering for a full stack Rails request (i.e. "ludicrous speed" pr) github.com/rails/rails/pu…
Removing object allocation sometimes give big perf gains. Sometimes it's small. Here's 4k fewer objects per request on pages that error github.com/rails/rails/pu…
Do you use url_for or know a friend who does? (Hint it's used under the hood all the time by other methods). Here's me making it 27% faster when called with a hash github.com/rails/rails/pu…
Oh how we love to check for nil? and empty? which is exctly what blank? does. This PR makes this very frequently called method about 30% faster github.com/rails/rails/pu…
How do you even go about increasing performance numbers? First you've got to have some good benchmarks. It's why I wrote derailed benchmarks github.com/schneems/derai…
One of my fave talks was about tuning Rails to use less memory and treating perf issues like regular bugs in "Speed Science"
If you're looking for faster asset generation, then look no further than "Slo Mo" that talks about how I made Sprockets significantly faster with lots of help from the great Ruby performance community
I don't just look at framework level perf issues. Many times the biggest offenders are in HOW we use a framework. I wrote extensively about these issues including reducing N+1 queries and memory schneems.com/2017/03/28/n1-…
Finding and fixing problematic queries schneems.com/2017/07/11/usi…
How I reduced my DB server load by 80% schneems.com/2017/07/18/how…
Speeding up random record lookup schneems.com/2017/11/29/say…
How the "worst performing web server" in Ruby is actually fairly performant schneems.com/2017/08/01/is-…
What to do when you see strange memory patterns in your Ruby apps schneems.com/2017/04/12/jum…
Patching Sprockets to get up to a 12x performance increase on secondary deploy schneems.com/blogs/2016-02-…
Working with heap dumps (granted, not that useful, but I wrote it anyway and it's focused on performance) schneems.com/2016/01/06/def…
How to debug a memory leak blog.codeship.com/debugging-a-me…
How another developer found a perf regresion in RubyGems and made it faster sitepoint.com/rubygems-slow/
Why hashie can cause performance issues and how I fixed some of them in omniauth schneems.com/2014/12/15/has…
That time when I identified that the mail gem was using 36 megs of Ram and helped the team identify the issue and fix it schneems.com/2014/11/07/i-r…
The zen and methods of fixing any performance issue blog.codeship.com/goal-performan…
How string keys in Hashes got way faster in Ruby 2.2 sitepoint.com/unraveling-str…
So nobody cares about performance in Rails @codinghorror? I care. Lots of people care.
