As you may be aware, the reading team was unhappy with its current release process and shared its rationale https://www.mediawiki.org/wiki/Reading/Web/Release_process/experiment#Rationale for a different release process. This resulted in an experiment on Gather, QuickSurveys, Cards and RelatedArticles where the team developed on the dev branch by default. When the team was happy that the dev branch was stable, it would tag a release and merge to master. The idea was to minimise the deployment of incomplete features and SWAT deploys.
This process saw mixed success. Although we felt more comfortable and in control of the code we shipped due to more time for designers and product owners to input into our process it was clear there was an impedance mismatch with the expectations in the MediaWiki developer community and for a small team of 5, we were unable to keep up with the periodic releases of extensions we were not actively maintaining.
As a result we are abandoning this experiment and going back to the Mediawiki Way™ of releasing and merging directly to master.
When we work on new features, we will be more mindful of +2, marking the initial commit of a chain of commits that requires sign off from a designer and product owner with a -2.
We think this will be a better approach for everyone. Our only remaining question is when and if to do release number bumps in future. We'll be back with an answer about that shortly... ideas welcomed.
Thank you for your patience while we experimented and let us know if there is anything else of interest to you in this experiment we have run.
Notes from post-mortem are on wiki https://www.mediawiki.org/wiki/Reading/Web/Notes/2016-Q3_Branching_strategy_post-mortem
wikitech-l@lists.wikimedia.org