From MozillaWiki
Jump to: navigation, search
  • To refresh on the list of Features and their priority, click here.
  • To refresh on the plan of record per Toronto's offsite, click here.
  • To see baseline vs. current reality, click here: here

ADU update

  • Aurora - 1,229 (+12%)
  • Nightly - 1,196 (+4%)
  • Beta - 27,042 (+3%)


To see who is attending the showcase


  • Dec 2011: Feature Complete


  • Jan 16 2012: String Freeze ==> Jan 30th (lock-step with cutover)
  • Jan 16 2012: First Run Walk-Through ==> needs to be reset; pending profile migration which is getting closer
  • Jan 30 2012: Beta Cutover & Beta Declare ==> merge happened but we are day-for day slip until we ship to the store
  • Feb 27 2012: Public Announcement at Mobile World Congress
  • March 30 2012: Final Release

Beta Readiness

  • Status:
  • We are string frozen: the focus is now to focus in page rendering (panning and zooming perf, quick/smooth tab switching and closing, awesome bar/screen responsiveness (esp. after syncing bookmarks and history)

We are in day-for-day slip for shipping Native Fennec Fx11 to the android market here's why

  • Tab: display tabs tray menu
  • Awesome Bar Perf after sync
  • Panning and zooming perf: we need to make a go/no go? We should probably ship tile-by-tile
  • Full AMO Sign-Off
  • Thumbnail Polish
  • Session Restore
  • Remaining sync blockers related to data integrity

Full list of blockers, there are some fixed ones in here because we want to be sure to qualify those fixes before beta:

Release Management

  • Should we go ahead and release XUL for both phones and tablets since we are not ready with Java?
  • What's the testing strategy for Beta? Testing the weekly build out of RelEng is likely too late given the amount of code change we expect to still take.
  • Is the new copy ready for native on beta?
  • Split-APK
    • Status: Waiting for the fix for bug 719560- this was fixed on m-c but backed out
    • Requires: QA testing on tablets/phones once posted
    • Fallback options: Push XUL to all for the time being, only push to tablets, hold off on pushing altogether
  • Profile migration
    • Status: QA should be able to provide testing status
  • Single-locale APKs
    • Won't be ready in time for 1/31 - focused on getting this onto m-c and aurora prior to uplifting to beta
    • Are multi-locale APKs still our our fallback?
    • Requires: QA test plan and verification in conjunction with split-APK work
    • Remaining multi-locale bug: bug 713464 (needs to be uplifted to aurora/beta)
  • L10N
    • Required for String Freeze for Fx11 (Jan 30th): query


    • Flash is working on ICS
    • First Run for current XUL users (profile migration)

Other Top Issues and Their Status


  • They are uplifted to Aurora: plan of record is to go to beta with history and bookmarks; no passwords, no tabs
  • We need to get to the bottom of awesome bar performance after sync
  • Password content provider bug 704682


Of our UX list for beta:

  • Done
    • Get the in-browser Sync UI pieces in, so people can find and set up sync bug 708414, bug 716906
    • Get the tab menu not covering the whole screen -- may seem minor, but I want people to get the right idea of how it works as of beta - bug 712526
  • Improved
    • aweseomescreen - slow top sites list coming in
    • tab menu - opening a tab, switching tabs, deleting tabs
    • Fix the p1/p2 font inflation bugs -- we want this to be as right as possible for beta; bug 706193, bug 706198
      • mobile sites exception is done - is it in nightlies/aurora?
    • Some thumbnail tweaks around differences in tab menu / start page, scaling where we shouldn't be (bug 721032)
  • Still to be fixed
    • Cut down the amount of checkerboarding - bug 716581
    • Get frecency / top sites as right as we can: bug 704977
    • Handful of startup weirdness bugs: tracked in bug 721008
    • [new] Overwhelming flat list of desktop bookmarks bug 716918

Website Compatibility

(Two-pronged approach)

MTD Update

MTD feedback in order of frequency:

  • "Feeling zippier"
  • Sync <--- people still want sync
  • Remaining checkerboarding
  • Localizations
  • Unzooming during page load
  • Frame rate when scrolling
  • Top sites still don't work
  • Flash on ICS


Show Case : https://wiki.mozilla.org/Fennec/NativeUI/Showcase2-1 

- Demo first then crazy stuff to talk about.
- Chang is moderating; Naoki taking notes

- talk about beta readiness; but before that Demos
- Snorp got flash on ICS ; will be in tomorrow's nightly
- www.homestarrunner.com demoed ; bad bandwidth
	- retry.
	- runs.
	- panned/zoomed/rotated ; much better experience than on gingerbread
	- refresh and crashed after rotation
	- does mean we support for all platform that support flash

- XUL to Native Migration demoed : 
	- first run from Xul demoed; set of default bookmarks and history
	- all that stuff should be migrated into native.
	- "Setting up Nightly" moving over bookmarks, history, passwords... etc.
	- the setting up nightly is shown because it affects UX too much
	- for a 50 meg places database 45 seconds or so...
		- some ideas on how to cut that down; but not yet
		- XUL bookmarks to Native bookmarks
		- single install for a single message 
		- only see this if you have a profile

- ADU update: climbing on the Aurora channel; 27k testers on beta
- schedule : trains are running on time
<read wiki for beta readiness>

- Release Management
	- can't ship native by friday; we need to make a decision either to respin or wait til beta 2
	- need to approve aurora, beta fixes.
	- because desktop is going out regardless, wait for native to have respin?  or push out xul, or push out xul to just tablets, or we can skip and wait til beta 2.
	- firefox 10 was the last XUL for beta?  What's the input from Jay?  Jay agrees with jonath
	- not concerned about XUL side of things; release yes, beta, no
	- go to build is typically tuesday evening, goes out friday.
	- testing plan for beta?
	- need to know the set of changes that are likely to land.
	- we might have regressions that may end up slipping; depends on QA/Releng
	- just wanted to call it out; probably not right meeting to talk about it in detail
	- most of the stuff on aurora, we should be uplifting to beta.
	- compare the change set of what's on aurora versus beta
	- is the checkerboarding ok?
		- huge regression from XUL
		- there are a lot of other things that are good
	- Panning slowly is frustrating
	- reasonable planning of a news article
		- double tap zoom in and double tap zoom out
	- wouldn't use it all the time currently because it doesn't do this reasonably
	- Jay getting a lot of checkerboarding; Jonath doesn't
	- Jonath is on aurora.  Jay is on trunk
	- we should look at beta
	- Alex is said is true up to yesterday.  Technically we weren't pushing everything fast as possible to aurora because of that.
	- as of yesterday it is now beta.  tiling should now be on aurora
	- evaluate checkerboarding on beta versus aurora
		- we don't have beta builds.  We have tinderbox builds
	- how's panning?
		- it's true that the migration happened
		- if we switch to aurora, you can test that another day
		- checkerboarding is a non-issue
		- double tap zoom in/out
		- we know that you get checkerboarding
		- we have plans to fix that up (Java compositor better, etc.)
		- main question : is it ok to ship with what we have on Beta?
	- Checkpoint tomorrow?
		- send a link : Erin's action item
		- releng to create a build?
		- candidate build for us
		- tinderbox build would be a bit crazy to deal with
	- the value for mozilla and the beta program; we need to get the data from the beta crowd
	- not enough data from crashes, native sync, awesome bar changes
	- want more people to use the product, so we can fix more bugs
	- concern : we don't want the beta people to walk out
	- are we ready for beta?
		- mfinkle/elancaster : not quite there; checkerboarding might be good enough, but not the other stuff
	- jay: "erin's list the same as mfinkle's list?" mfinkle: "yes."
	- session restore we wanted in - start up user experience
	- fixed other ui stuff; restore issues fixed
	- more work to be done
	- need to look at startup ux; to make sure it's good enough for beta
	- tab/awesome bar experience 
	- polish around thumbnails
	- pretty good on ICS; notice that there's some aspect ratio nonsense going on
	- responsiveness and minor polish things to make us look better
	- finding, installing, uninstalling addons?
		- AMO now has support for Native UI addons
		- QA can now start doing that
		- string freeze now; if there's an issues such as an ad for AMO, then we would need to break string freeze
		- testing addons; need to make sure that is working (few people have been doing that)
		- we need to test from AMO for UX
	- once we get these bugs fixed, and group together to eval checkerboarding, would we ship?  blockers down to zero?
		- it might still be missing a couple of things such as Sync (passwords are going to make it or not?)
		- some of the things that mark mentioned such as polish issues that we can kick out
		- can tune the list and kick out
		- brad, mark, doug should go through the list to trim the fat : http://bit.ly/yrhZ2a
		- don't have password content providers
		- fix existing blockers
		- Jay : "password sync is a ship blocker.  not a beta 1 blocker, but it should come soon"
		- Tough to do a demo in mobile world congress without it
	- one section for mobile, for everything else : desktop vs mobile bookmarks.  we would have to break string freeze for it.
		- we can create sections so that they aren't intermingled.
	- go to beta 1, bookmarks, history no password/tabs  beta 2/3 should have passwords
		- history in bookmarks crash
		- tag sync blocker bugs with the keyword tag
		- 2 or 3,... not 5 or 10 blockers
	- comfortable with the list
- How do we get to burn down?  When do we get it ready?
- Profile migration testing has been done by QA
	- just need to talk to GCP to make sure expectations are right
	- backed out of the split-APK.  ndon't need to talk about that
	- startup metric need to be discussed for multi/single
		- need to decide
		~ 400 ms between single / multi locale
	- mark/brad/doug : release single locale builds was the thought
		- one of the goals was startup performance for everythign
		- doug : doesn't want to take .5 hit for multi locale
	- aki wasn't able to give hard time line
		- jay gives a thumbs up
	- alex to go through with single locale builds : hard blocker now
	- maybe in future or others, we will go with multilocale (for amazon)
	- google supports single locale; we should go with that.
	- we should think about amazon separately
		- not all the work for single locale is in.  it will need to be uplifted all the way up
		- still a good few days; calling that out.
		- may slip to beta 3 or out of band beta
		- do we want to gate on the specific locale?
		- beta 1 or final?  first beta can be pretty limited - en-US should be ok
		- we want the right solution
- sounds like we have a baseline for beta blocker list
	- doing an analysis and figuring out what will fit and when
	- we need 10th for in the store, 8th/9th for releng
	- need to take a look at the list in order to figure out if 7th is doable
	- need readability dev to talk about those bugs
	- header for inflation is good enough for beta, but not for ship
	- awesomeness calc ... need to spend some time now
	- tabs took too long, we made improvements

- most of the bugs on the list can be broken down to 4 or 5... so it's do able.
- if they are not well owned, they will be.
	- 7th should be doable as a drop dead date.

- do we have checkerboarding well enough, and then push it.
	- take a look at some of the other ones on the list.	
	- check point should be tomorrow.
- publish bug is of concern : 719560
	- checkerboard fixes is going to be more risky and harder than before
- to be clear : if checkerboarding is ok for beta, we would want to make it better, and scary changes will be being pushed to beta
- sprint to get some form of gllayers up on native
	- not production ready but standing up sprint for the week

- the top issue for final ship
- why does gllayer block ship?
	- a lot of these system we don't know how they perform unless you stand them up
	- java compositor is ok, but not great; better than software
	- tile by tile
	- other incremental improvements... can't do great
	- we didn't know how long it would take for these improvements
	- open gl layers is done
	- async panning is done
	- integration needs to be done now
	- high risk, but good rewards for panning.
GLLayer info:
[1:04pm] jprmc: done for Fennec XUL back in the fall
[1:04pm] jprmc: does not work with architecture in Fennec native
[1:05pm] jprmc: this is high risk
[1:05pm] jprmc: and unclear what the perf is yet

- checkerboarding check point
- caveat with the checkerboarding builds; uninstall the nightly and install the build
- we only discussed beta.  checkpoint tomorrow.
- we don't know until we stand it up what the risks are.
- make sure that bob, aki are there.