QA/SoftVision/Meetings/Desktop:2014-04-02

From MozillaWiki
< QA‎ | SoftVision‎ | Meetings
Jump to: navigation, search

Desktop QA Standup - April 2nd, 2014

Dial-in Information

Attendees: Manuela, Florin, Tracy, Camelia, Bogdan, Ioana, Alexandra, Nils, Cornel, Andrei, Mihaela, Petruta, KaiRo, Otilia, Matt, Anthony Hughes

Heads-Up

  • Upcoming PTOs:
    • Liz - PyCon next week Apr 8-15
  • Events
    • Next week QA community champions - in SF,

Upcoming Schedule

Previous Action Items

  • mschifer to follow up with Platform/Automation re: WebGL automation coverage (so we can catch issues like bug 975824 more easily)
    • WebGL test suties available, not kept up/disabled, Marc to talk to them about how we can help them.
  • Safe browsing: Matt to investigate history of feature, what goes into Fx29, what is in test suite, how to cover in test day
  • Where are we with keywords and qa whiteboard tags.
    • [Kairo] Send out a summary to dev-quality and dev-planning and elicit discussion.
  • Think about focusing our triage process; look at bugs that have been tagged by developers in platform, but ignore otherwise - proposal mentioned last time: any bugs out of the Firefox iteration list + anything marked qa+ or qa? (or verifyme?) in Firefox/Toolkit/Core products (and marked with the appropriate target milestone)
    • We currently triage Firefox, Core, Toolkit for incoming bugs and broken down by release.
    • What do people think about that?
    • Until the whole Firefox desktop team is on iterations, will actually have to have first part as "any bugs fixed in iteration or Firefox product marked with that target milestone"
    • We usually already miss most of the Toolkit/Core bugs and the vast majority of those would be qa- so it's probably better to not triage all of those but concentrate on those that get flagged as needing us.
    • If we go with that, we need to heavily communicate this to developers so qa+/? actually get set where needed
    • [ashughes] I think I prefer to use verifyme in place of [qa+] and continue to use [qa-], don't find [qa!] all that valuable -- this is largely what we're doing outside of iterations (ie. Platform) so it might be the least amount "friction"
    • [lizzard] I'd like to use [verify+] [verify-] [verify?] for verifying, and [qa+-?] for general requests for qa help at any stage. testday is a good tag. I don't love the testday-YYMMDD tags and I think we should stop using them. closeme and dupeme are useful. I'd also like to start constructing queries that exclude/include needinfo, since we have a lot of incoming bugs that are waiting on an answer from the bug reporter. Bugs with *nothing* in the qa whiteboard and no needinfo flag, I would put into the "incoming" triage bucket; it's useful to know what hasn't been looked at by anyone yet vs. what's in process.
    • Existing equivalences:

qa+ | verifyme, qawanted qa? | qawanted qa- | keyword needed qa! | lack of qa keywords in the bug report

Discussion Items

  • [Florin] How should we handle bugs Resolved in sprint (NOT triaged and with no qa contact)? - seems we can get Resolved bugs early on in the Sprint, and we may want to avoid getting a large list of fixes to work on towards the end of the sprint.
  • [Florin] Ioana being replaced by Florin Mezei - Should add Florin in all places where Ioana is currently added and remove Ioana from them
    • Release-drivers, qa-drivers, qa-contribute, desktopqa
    • Security bugs
    • Any other places
  • Fx28 safe browsing hotfix
    • Spec: https://wiki.mozilla.org/Security/Features/Application_Reputation_Design_Doc
    • I had vague knowledge of a change for Fx28 but did not follow up
      • Feature was to be limited in scope for Fx28
      • Initially we just wanted to download new lists of blacklisted URLs etc. from Google but not use them
      • We ended up downloading them and using them in some capacity anyway
    • Feature was working as designed, and we would not have caught during testing
    • Post-launch, Google alerted us that we generated too much traffic, more than expected
    • We realized that our design was faulty - too many unnecessary lookups
    • We turned off logic for this feature via hotfix preference change
    • Fx29 will use this feature partially
    • Fx30 will use this feature fully
    • Test exists in MozMill, but we're working with Henrik to make another test that would have caught the above specific issue

Channel Status

Beta (Fx 29)

Aurora (Fx 30)

Nightly (Fx 31)

  • mozilla::pkix SSL cert validation library landed

ESR (Fx 24.4)

Security

  • SSL compat, root cert removal

Automation

  • Goals have been set for Q2: https://etherpad.mozilla.org/firefox-automation-goals
  • TPS
    • Good progress, only two failing tests are remaining (bug 981706)
    • We will bring back the CI hopefully today
  • Mozmill Tests
    • Lots of regressions (SSL, video) and sporadic crashes this week
    • Working with mmc on possible safe browsing tests for Mozmill

Stability

Community

  • I shared adalucinet's awesome contributor spreadsheet with other community champions. Thanks ada!
  • We have 330+ contributors for q1 across most of the QA teams. They are not broken out by week of active contribution but we are moving closer to tracking them. One possibility is emailing all 300+ of the known 2014 contributors to invite them to make a one and done account.
  • lizzard will have an intern from the GNOME-OPW project for summer, interested in developing tools for community contribution (probably one and done)

Notes and Actions

  • [juanb, ashughes] Add Florin to relevant mailing lists.
  • [mschifer] review KaiRo's dev-quality post