Auto-tools/Goals/YearEnd2011

From MozillaWiki
Jump to: navigation, search

This year has been a rough one for the Mozilla Project. We've grown quite a bit and with that growth came several challenges and opportunities as the stress on every single system increased. The A*Team was in the center of many of these issues and we did extremely well with the curveballs and reverses we were thrown.

We rose to every challenge, and each time we left the automation systems we touched in a better position for both their current and future needs. This has all come together very well and has positioned us for a stellar 2012. Before talking about 2012 though, let's roll the clock back and look at some of the highlights.

In January of 2011... In December of 2011...
We could not get Android automation deployed Android XUL automation > 95% green
Bughunter had melted down Couch, we were scrambling to rebuild it
  • Bughunter keeps up with daily crash load
  • Bughunter adapted to rapid release cycle
  • Bughunter's new UI ready to launch
No Tp5, Tspaint, xperf, Android RSS measurements All of that has landed
No responsiveness testing at all
  • Talos responsiveness designed, implemented, and landed
  • Peptest designed, implemented, and landed
It took 6-8 hours from checkin to tests done It's now 2-3.5 hours
SpecialPowers was barely integrated into mochitest
  • Now fully integrated into mochitests
  • Developers using it when they write tests
No such thing as B2G We now have automation for B2G
No such thing as Android Native Fennec We almost have all Android automation rebuilt for native Fennec
Templeton, Autolog were "Big Dreams" They're now deployed and cornerstones to many automation projects
Mozbase/Mozharness existed only on a whiteboard It's now the system behind peptest deployment and soon to be rolled out in talos.
Eideticker was a pie-in-the-sky idea It's almost in production
Orange Factor was a new freshly minted website and we didn't now if the WOO would work WOO worked enabling us to change how we land code at Mozilla, paving the way for mozilla-inbound
Bugzilla had no splinter, rest API, extensions, or future planning It does now.

This year we added dkl, glob, wlach, mdas, and jeads to the team. We all got better at blogging, and the resulting increase in our visibility brought us two new active contributors: Byk and Mavericks.

With our work in 2011 we created a great set of building blocks for future projects. Think about mozbase, eideticker, mozharness, bughunter, templeton, autolog, marionette, our changes to bugzilla, and you can start to see how all the extensibility we built into our tools will enable us to put these building blocks together in many different ways to solve a wide array of problems.

So, for 2012, I envision us doing exactly that--tailoring our projects, completing the work we've started, and using these extension points to make our automation systems address the needs of the Mozilla Project as it moves toward new frontiers like Apps, Identity, B2G, and beyond.

I don't have an exact number of hours we have saved developers by the resulting increases in productivity that we've made possible (and the corresponding decreases in frustration), but I know it was a upwards of thousands of hours.

The impact we've had at Mozilla is very real, while our developers may get glory for building the open web, we construct the very scaffolding they stand on. I'm proud to be one of you, and I'm proud of each of you for the way that you've come together, leading on our team and across teams to address challenges we all faced. I'm looking forward to 2012.

Roadmap for 2012