forked from servo/servo
-
Notifications
You must be signed in to change notification settings - Fork 0
Meeting 2014 04 21
Josh Matthews edited this page Apr 21, 2014
·
1 revision
Please update your status at: http://benjamin.smedbergs.us/weekly-updates.fcgi/
- Rust upgrade status (larsberg)
- mach (jack)
- Some larger work items https://etherpad.mozilla.org/Servo-samsung-tasks (larsberg)
- lars: building on ms2ger's PR, I have all submodules building and I'm into the main crate. The biggest thing right now is what we expected: changes to visibility. Because everything became private all the submodules around C apis all got switched to priv and I'm doing the tedious process of making them pub one at a time. Neither ms2ger or I have rebased it in the last week and a half, so we may get a big surprise there.
- jack: should we wait to rebase rooting until after upgrade?
- jdm: that seems reasonable.
- simon: rebase on top of servo master or rust master?
- jack: servo.
- jack: Between meetings, I looked into mach and got it set up (despite the lack of documentation). We should be able to make it work with Servo. I have a local branch that has mach-in-tree and a command that tests it out. The question is: it can handle all of our odd jobs and it's easy to add them, but what should the first command be? Not just hello, world, but something that we'd use.
- jdm: web-platform-tests!
- mbrubeck: Yes, tests. "mach ref-test "
- jdm: There's good integration with mach for w-p-t that jgraham knows about, so we should steal that.
- jack: I will ping him and take a look at that. I didn't want to start with build because we will have to write that from scratch (cannot reuse Firefox's complicated one). I will start with that and see if I can get one of the test suites running through mach. After that, we can all hack away on it. All it does is load files with the commands and command argument parsing. It does look like it'll fit in nicely for removing our submodule checkout hacks in configure. That will be nice. I also want to get in the rust compiler stuff at some point, so that we can skip it, etc.
- lars: I just wanted to mention these to the group. We're looking for tasks that seem like they'd be easier to hand off to Samsung. Large and meaty tasks that aren't quite as invovled as major layout features. There are some here, but if you have more ideas, just add them.
- jack: They seemed interested in Web Components. We're not 100% certain how to do that, but we have some paths around web components that don't require doing that inside of Servo (to avoid the architectural work).
- ms2ger: Status of HTML parser
- jack: He's optimizing the tokenizer and moving along. We can get a status update and put it here.
- jack: I'll probably do a walkthrough for our new intern and one new hire, both of whom are starting next week.
- jack: Try filing the web-platform-tests-related bugs! They're easy ones for new contributors.
- manishearth: https://github.com/mozilla/servo/wiki/Running-Web-Platform-Tests-on-Servo
- jack: We've had lots of new contributors lately.
- jdm: Got ACID2 ones
- mbrubeck: cjones is looking at Servo in rr
- pcwalton: I put a giant review in the queue
- larsberg: I'll look at it shortly and SimonSapin will also look at it on Tuesday when UK is off of vacation.
- pcwalton: Lots of PRs in the queue. Is there a timing for the new rooting?
- jack: I'm reviewing it today. Will take a couple of days - it's enormous. And jdm keeps adding another thousand lines here and there.
- jdm: It's OVER 9000!
- pcwalton: It's great work.
- jack: At least two days.
- pcwalton: Is it foolproof?
- jdm: Yes. Some things can be slotted in that aren't foolproof at the moment.
- pcwalton: Sounds similar to layout. Foolproof design, but a couple of places.
- mbrubeck: I've been pulling at the iframe thread for a few days now. Does anyone know when they were last working? I don't want to git rebase across upgrades...
- jack: Early october for sure (the summit demos had parallel iframes). I assume they continued to work until a horrible rust upgrade from early december to early January.
- mbrubeck: I've traced some fo the problems to a recent upgrade, but some seem much older. Some were curious as to whether they ever worked...
- jack: I'm surprised there's no iframe reftest.
- mbrubeck: It was checked in commented-out, due to "crashiness."
- jack: Shame on us! Once you find it, hopefulyl we can uncomment the reftest.
- mbrubeck: Definitely.
- jack: Maybe bors should uncomment them after a month...
- mbrubeck: Or crowbot can nag you!
- jack: I assume iframes will still not work, even after you track it down, due to scissoring?
- mbrubeck: Just tryign to get them to not crash and paint them.
- pcwalton: If they're small enough not to require scrolling, they should work correctly. Shouldn't be hard to add, but it will change layers and will have to be completely rewritten.
- jack: Can also use overflow:hidden on the iframe, right?
- pcwalton: Yes, it's a totally separate codepath from iframe stuff (Because it's layerized). So the render task does the clipping for overflow:hidden, but the compositor does it for iframes.
- jack: Fine. A small iframe test case would be great to just get it up and running again.
- pcwalton: Yes, fixing the pipeline issues is key. I don't have any plans for rewriting that right now.
- jack: Probably revisit when we get browser context support.
- pcwalton: We'll have to have it fully operational before we can throw it away, because otherwise you don't get all of the constraints quite right.