On 05/08/2014 01:27 AM, Sumana Harihareswara wrote:
I think I've articulated our key performance principles, and am looking for good and bad examples. Our performance guidelines will be a set of values/principles, each one with a good example and a bad example drawn from our own experience. Here's the list:
https://www.mediawiki.org/wiki/Performance_guidelines#General_performance_pr...
Please feel free to paste links on the talk page, saying whether it's a good or bad example - I will add prose and explanation. I'll also be gathering examples on Friday at the Zurich hackathon. I want to polish this and get it approved during the hackathon.
(I'm still cleaning up the detailed explanation of each principle.)
RfC we can use to discuss larger questions: https://www.mediawiki.org/wiki/Requests_for_comment/Performance_standards_fo...
These guidelines are a lot more readable and helpful now, but I still need help with good and bad examples for:
Choose the right persistence layer for your needs: Redis job queue, MariaDB database, Swift file store, or memcached cache. Wikimedia uses and depends heavily on many different caching layers, so your code needs to work in that environment! (But it also must work if everything misses cache.)
And I need the *code* for good and bad examples of:
Users should have a smooth experience; different components should render progressively. Preserve positioning of elements (e.g. avoid pushing content in a reflow).