This PR also took the opportunity to hand-optimize and hand-make retina variants of all images in the blog and lifestream. This eliminated most of the large asset warnings from #webpack #performance #hypstatic

Optimized String#scan in ferrocarril to be comparable to CRuby performance #rust #ruby #mruby #performance https://github.com/artichoke/ferrocarril/pull/138

it looks like prod peaks at 75 req/s #performance

OMG just rewrote my #git PS1 which runs as part of my PROMPT_COMMAND. Now 20% faster both inside and outside a repo. terminal feels so much faster now #win #performance https://github.com/lopopolo/dotfiles/commit/2d5054c4ce

DOMContentLoaded in less than 200ms on lifestream index #performance #cdn

Switching from yui-compressor to yuglify saved me 1KB on bootstrap.js and 30 bytes on bootstrap+hyperbola css #performance

lifestream template refactor didn't yield any #performance wins, but the code is much cleaner now. #win

Front page warm nginx latency is 12ms; contact page is 20ms #performance

Current prod gunicorn time (measured at nginx) for the lifestream index page is 70ms with no partial caching and 40ms with the sidebar cached #performance

Tested rendering the lifestream index with all 3 interpreters. pypy is slower than python2.7 by 2x. python3.5 is faster than python2.7 by 2x. #performance

Sped up the lifestream page by 30ms by caching the archive sidebar. More template speedups to be had by switching to python3 (str to unicode coercion in python2) #win #performance

60ms server-side latency on the lifestream page now #performance

got another big #performance win by not minifying HTML in python not doing so shaved ~100ms off response time which means my server was spending more than 100ms of CPU time for the pleasure #fail

about ~20% of server time for lifestream is spent reversing URLs for hashtags :/ #performance #fail

Latency at nginx for hyperbola: frontpage: 25ms, contact: 40ms, lifestream: 200ms #performance

Switching gunicorn to a domain socket instead of a TCP port sped up the site by 2x #performance

Spent the last week optimizing, re-styling, and significantly refactoring hyperbola. The lifestream page loads ~300ms faster and DOMReady is ~600ms faster #performance

Get optimized, HTML! Just added some middleware that minifies Django template HTML output #performance Google pagespeed really likes my site now (on Desktops) ... 94/100 #win

Throughput on the #box #metadata backend is limited by the rate at which our load driver can make requests. 2ms latency at the client. This thing is #fast #performance

Woo: hyperbola, now with thumbnailing #performance