EngineeringProductivity/Meetings/2012-02-27
From MozillaWiki
< EngineeringProductivity | Meetings(Redirected from Auto-tools/Meetings/2012-02-27)
The Overview
- Bugzilla
- [dkl] More work on AutoLand extension. Extra requirements added on last minute and once testing on those are complete, then time to push off to a security review.
- [dkl] Upstream Bugzilla: Helped with a recent security release as well as the release of the new Bugzilla version 4.2. So now work to start on getting our BMO converted over to the new 4.2 version.
- [dkl] Upstream work on patches and reviews
- [dkl] Normal BMO Administrative work.
- [glob] qa migration work (dev.planning announcement, followup discussion)
- [glob] new group security report for infrasec
- [glob] bmo ui bug fixes (inline-history, splinter)
- [glob] lots of admin work (admin work covers things like adding new flags, components, products, versions, etc)
- [glob] upstream work (meetings, reviews, etc)
- Bughunter
- [bc] No users. Need to get new test vms in order to be able to promote the use I think. In the mean time, I've been filing in keeping track of bugs and such. Been testing several outstanding patches and almost ready to land them in the repo. Tomcat is bringing up a version of bughunter for use in top site testing etc.
- [jeads] Added some UI elements for highlighting crashes that are exploitable. Fixed some bugs. Not committed yet, should commit early this week.
- Marionette (https://developer.mozilla.org/en/Marionette)
- [mdas] added findElement/findElements support to chrome and content, and that involved some code reorganization. We also responded to the sec team's concern about preferences and changed how marionette gets enabled/disabled.
- [mdas] worked on recording touch events from the emulator and playing them back. Should have real touch support next quarter, but this is something that QA can use for tests in the meantime. It has yet to be committed, I'd like to get some feedback on it first.
- [mdas] working on general marionette implementation stuff
- [jgriffin] added tagging of async script calls, which will fix this bug: https://bugzilla.mozilla.org/show_bug.cgi?id=725754
- [jgriffin] refactored a lot of the code for the execute functions, in order to isolate them better so that they can't contaminate each other (e.g., a stray setTimeout from one execute function sending a result during a subsequent execute).
- [jgriffin] finished the Jenkins build and test system for b2g. One remaining item is to update the Marionette test runner to optionally output test results in JUnit XML, so that Jenkins can do interesting things with them. Currently it recognizes pass/fail status, but with the XML it will be able to display graphs over time, etc.
- [jgriffin] added a Jenkins auto-merge system for b2g, that auto-merges changes from m-c into the b2g gecko fork. This is running right now on a test fork and will be migrated to the real b2g fork soon.
- Signal From Noise
- [jhammel] Pageloader is now part of Talos
- [jhammel] we're getting close to be able to run talos without CWD being "special"
- [jhammel] working on separating out statistics from Talos and Pageloader which will hopefully be consumed by graphserver and a new package (https://bugzilla.mozilla.org/show_bug.cgi?id=728461)
- [jeads] Built a new graphs schema with jmaher, jhammel, and ctalbert. Working with jmaher to build a webservice that populates the new schema.
- [jeads] Adapted the bughunter data view concept to work with s4n data. There's a fully functional user interface pulling graph data... Doesn't do anything useful yet but it's a good first step.
- [jmaher] finished cleaning up side by side staging from previous weeks
- [jmaher] starting to make progress on defining a more reliable tp5 test run- experimentations are almost complete
- [jmaher] landed a lot more cleanup and configuration patches to talos
- Mobile Automation
- [mcote] Redesigned to load tests dynamically and run them in separate processes. Working on pulse support.
- [jmaher] filed bugs for failing reftests
- [jmaher] panda board is making progress, waiting on updated hardware (ES model)
- [jmaher] eta for robocop tests on talos is this week
- [jmaher] xpcshell for android is almost ready for staging
- [wlach] working on getting Android 4.0 running on the pandaboards with ctalbert and jmaher.
- Eideticker (https://wiki.mozilla.org/Project_Eideticker):
- [wlach] Some work done on making Eideticker tests work across different browsers (e.g. stock android, chrome)
- [wlach] Related to the above, finished an abstraction to start an android application, as well as a new devicemanager method to run a command in a shell and get its output (useful for running xpcshell tests on Android).
- Talos
- [wlach] worked on removing the pywin32 dependency from talos, which should pave the way to stop needing to support python 2.4 (which we've been using up to now because we don't have a more recent version of pywin32 on the build slaves).
- Ateam Services
- [bc] Waiting on network configuration changes.
- Jetperf Harness (http://k0s.org/mozilla/hg/jetperf)
- [jhammel] RelEng has weighed in on some structure issues (https://bugzilla.mozilla.org/show_bug.cgi?id=729205) that will require substantive modifications on the harness so a big step back there
- [jhammel] https://bugzilla.mozilla.org/show_bug.cgi?id=700722 will also be a blocker here, as it was for desktop talos + mozharness. i've unbitrotted the patch and will try it soon
- [jhammel] expecting to devote most of march to getting Jetperf up and running. These unexpected setbacks plus the need to integrate to buildbot, etc, will be my first priority.
- PyPI
- [jhammel] We need a python package index for internal installation. e.g. for releng purposes. Not a new project but i've been looking at this in this time period. I was originally going to use https://github.com/SurveyMonkey/CheesePrism but am seeing bugs. So I'm trying to figure out whether to spend time fixing that or developing a solution from scratch. I'm exploring both alternatives
- Peptest
- [mcote] Up on try (add &noignore=1 to the URL to see it; it's the "U"). Unsurprisingly, all oranges, since there are unresponsive periods in all the tests. There is quite a bit of variance in the results, though, so coming up with a baseline to use to catch regressions may be quite tricky. I am collecting stats to try to better understand what's happening.
Upcoming Events
All times Pacific Time, click on link to find your timezone
- Platform Meeting (all platform engineers) Tuesday at 11AM, Conf # 8605.
- Mobile Test Automation Wednesday at 10:30AM #304
- Marionette/Eideticker Meeting Thursday at 10AM Conf # 304
- Signal from Noise Thursday at 11AM #343
Round Table
- First pass at the graph schema from tuesday's meeting Media:Graphs_exp_v3141.pdf
- Networking Team needs someone to work with them to set up their NeckoAutomation
- goals status
Misc
In honor of everyone's hard work, please enjoy your Monday Unicorn Moment (MUM for short).