Representatives from these functional groups have been asked to
propose goals for their efforts for the second quarter of 2007:
Layout: dbaron and roc
GFX: Vlad and Stuart
Mac: Josh and Colin
Content: jst and sicking
JS: Brendan, Graydon
Moz2: Brendan (Moz2 may overlap with others)
XULRunner, Breakpad, Build System: Smedberg
Internationalization: Simon Montagu
Cycle Collector: Graydon
Meeting agenda, notes, and ongoing updates to the goals discussed in
this meeting can be found here:
Are there plans to record this meeting or have minutes taken? If not,
would it be appreciated if fantasai did so?
Platform Goals Meeting
Damon: Ok, first off let's talk about the overall process.
Damon: We have a wiki page here, we want to drop individual function
group goals into this wiki page
Damon: We want to focus more on functional goals/groups
Damon lists groups
Damon: Goals are aligned with Mitchell's suggestion for 18 month goals
Damon: We're focusing on 2007 Q2
Damon: We'll open this up for discussion on m.dev.platform
Damon: for a couple weeks
roc doesn't seem to be online
Damon: Enable new text frames is familiar
David: What I put on here is too much for one quarter, and I already
crossed some things off
David: One thing I crossed off is something roc wanted ..
David: for the high dpi stuff to be useful we need media queries (?)
David: And that's certainly not going to happen for 1.9
David: One goal is better testing
David: I want to make the reftests in for that before I forget
<Vlad> we need css3 content attribute in addition to media queries
David: I'd be interested in hearing how important other people thing
David: The bugs are a mix of regressions and new features
David: Some are not regressions in the 1.9 cycle, but longstanding bugs
we should be making progress on.
David: The mix on the list may not be the mix in terms of how I spend
Damon: If anyone has questions on these bugs, open this up on dev.platform
Vlad: I've be rewriting the css rendering to use thebes directly ...
Simon seems not to be online either
Second issue is too vague.
Does that mean we'll have no issues with i18n on textframes?
Boris: I don't really know how much time I'll end up having, so I'm being
Bors: These are things I'm trying to work on now.
Boris: I've been sitting on a big review since Sept and one of my goals
is to actually do it.
Boris: Single line text controls is something I had a patch for at some
Boris: but it doesn't quite work because of <br> issue in editor
Boris: On pages with single line text controls, that will save us a lot
Boris: right now we turn on the editor for text controls even when they
bsmedberg: Are there things people can help with the pop-up rewrite?
?: Can we generate automated text cases?
Boris: Neil has been working on automated UI testing stuff, we should
Boris: I don't think having the test cases will help with review, but
would help with confidence about landing it.
Enn: I have more tests, but they're not automated
Vlad: Any capabilities anyone wants for the graphics platform should be
specified by the end of the quarter.
Vlad: ? might be a cross-goal between us and firefox
Vlad: We're very confident about it (?) on windows and mac, some of us
confident on linux
Vlad: We're hoping to have a decent handle on that at the end of the quarter
Vlad: in order of confidence, it's win, mac, linux
Vlad: Nothing's rendering oddly..
Vlad: some of this runs into cssrendering
Vlad: The third thing here was also vague
Vlad: we should add an overall goal about performance
Vlad: a lot of it isn't gfx directly
Vlad: for this quarter we should categorize perf losses, not necessarily
fix them but figure out where we have problems
Stuart: the graph server that ... definite performance regressions
Vlad: especially for i18n text
stuart: ... glyphs for which fonts aren't installed
Vlad: take a hard look at our international page rendering performance
Vlad: what fonts are installed, whether they make sense
Vlad: any correctness problems..
Boris: As long as we're looking at tests, do we want to re-evaluate the
numbers we're producing on tinderbox?
Alice has a test box that has better page sets, more international pages
Boris: Should that be a goal for this quarter or next quarter? I lean
towards this quarter.
Vlad: I'd like us to understand what our performance difference is from
1.8 in this quarter
Vlad: And focus on fixing it next quarter
Sicking: Can we get help from QA with that?
Sicking: something that would be really good is to do ? automatically
Sicking: get a binary search or tree hash or graph
Vlad: we talked about the past 18-24 months taking builds of 1.9 and
1.8 branch builds each month and comparing
Vlad: we have the builds available
Vlad: should do that automatically
Mac OS X
Damon: One of our big goals is to stabilize cocoa
Damon: we're having a big hackathon this month
Damon: I'll let Josh speak
Josh: We've got 30 bugs or so.. 4-5 have patches waiting for review
Josh: working on drag and drop
Josh: cbarrett is working on scrollbars, so we'll have scrollbars clipping etc
Josh: I'm working on focus issues
Damon?: We're also focusing on .. event loop bug that is blocking some others
?: Questions on mac component widgets?
<Vlad> bsmedberg: --enable-libxul is going to break my canvas3d stuff so
hard, I bet!
<Vlad> though I think I can work around most breakage
<bsmedberg> Vlad: yeah, without help it will... let's grab an hour or
two next week
<Vlad> sounds good
Boris: I listed things I'm actively working on
Boris: .. text frames ..
Boris: editor is something we needed to do a long time, hoping to work
on it this quarter, depends on how text frame goes
Boris: some legwork on text frames needed to make this work without perf
Sicking: Would be great if we could stop adding <br> elements
Sicking: What I'm worried about is unknown consequences
Sicking: since we'll do a lot of things differently
Sicking: we could get bad regressions, e.g. with caret placement
Boris: basic problem is that this block ? initializing editor
Boris: I have done fair amount of testing with caret movement, seems to
Boris: main concerns are performance.
Boris: selection is painful, spellcheck can freeze the browser
Boris: This is something I'm hoping to get done this quarter, but it
might not happen
Johnny: First I'm hearing about it. Do we have to do it for 1.9?
Boris: It's nice for page layout. We don't /have/ to do it.
David: You're only talking about plain text?
Boris: Yes, just plain text. I'm not touching the html editor
David: There's a branch in 2001 branch .. mailnews performance branch
.. that removes <br> from html editor
Sicking: So maybe we need to get layout to be smarter about which lines
need to be relowed?
Boris: exactly, roc and I will be working on that
* fantasai can't hear
Damon: JS is next, graydon is not here but marked goals in wiki
Damon: Questions on that on platform list
David: X is going to fix the standard reference implementation of JS 2
to pass Y
?: I don't want to give other people goals here, but there's a security
issue with ???
will add bug #, but not description to notes
Benjamin: it's a cross-team goal with breakpad team to get breakpad to
do things .. needs to do for release
Benjamin: FF and build system
Benjamin: not clear whether we'll do xulrunner for FF3, but we'd like
that system running to support Mozilla 2 work
Benjamin: It produces builds, but you can't ship builds to something else
Benjamin: enabling xul.. I'm hopeful about this
Benjamin: It has potential to increase performance , but also has potential
to decrease performance on linux
<gnubeard> stuart talking
Benjamin: I hope to get libxul turned on for the next cycle(?)
Benjamin: when it lands is mostly irrelevant, it's for supporting Mozilla 2
puzzled looks around the table
Josh: I'm still confused about this. Is it going to change installation
Josh: Are you going to install libxul in a system directory, in an
Benjamin: The user won't see it. Not in the system directory.
Stuart: Do we have an estimate when we're using libxul?
Benjamin: If it doesn't happen by alpha 5 we aren't doing it
Damon: We can continue this conversation in dev.platform
Damon: create a concrete timeline for system reworking and replacement
Benjamin: We'd like to modularize the system so we can build multiple
apps on top of the core, so people don't have to fuss with
Benjamin: We might be switching to python for that, but we're not sure
Damon: Any additional goals we need to add?
Sicking: I have a question about timeline about final release
Damon: We're aiming for a beta by mid-July
Josh: safely before the end of the year
Damon: Let's focus these goals on the next three months: where do we want
to be in the next three months?
Damon: we should have a significant reduction in blocker bugs, then we
look at where we are at the end of the three months
Damon: These goals are specifically for this quarter. Aiming for a beta
Damon: Talk about it in .dev.platform
Damon: I'll send a note about next meeting to talk about goals again.
Damon: any other questions?
Boris: As far as scheduling goes..
Boris: The main categories we're concerned with now are blockers and
Boris: We're working through nominations dumping in two buckets, or minusing
Boris: For blocker bugs, we need to start looking for people to fix them
Boris: We're have hundreds of bugs that are either blocking or wanted
David: We need to start focusing on these + bugs now, or three months ago
Stuart: We're going to have our hands full just with the + bugs
Sicking: The wanted bugs are the ones that we want, but the people in this
room probably won't get to
?: They're things if someone volunteers we'll point them to as something
we really want.
Vlad: If we fixed the blocked bugs, those are the ones we need for a
Vlad: The wanted bugs are just bonus
Stuart: If they'll block a release, they're blockers. If they won't, then
Stuart: It's too late to be aiming for new features. Big regressions
should be plussed though
Damon: any other comments?