emilio: Hey, around?

nox: back from PTO?

nox: just in time to review #21029 :)

nox: omw to the office, will be in a couple minutes

ajeffrey|uk: Looking.

emilio: I had ideas about how to encode C enums in bindgen.

emilio: The gist is #[repr(transparent)] struct SomeCEnum(pub i32); const SomeVariant: SomeCEnum = SomeCEnum(32);

nox: thanks!

emilio: This brings us a new type for the enum, ability to discriminate it against known enum values, being able to represent values that are not known enum values, and finally no weird namespacing.

emilio: Tell me what you think when you have some time.

ajeffrey|uk: Are you telling me no one reviewed your PR in 2 damn weeks? :|

ajeffrey|uk: Two commits need to be squashed btw.

nox: we've been fighting CI the whole time.

nox: that sounds great. I think you still need the weird `SomeVariant_SomeCEnum` name (or associated consts), to handle `enum class` and co without name clashing

emilio: Yes, associated consts for C++ enum classes,

nox: but generally looks good, and should be easy. I think bindgen should do associated consts by default now

emilio: and normal consts for C ones.

nox: moving the arm builds to clang, and dealing with old MSVC builders.

All of this stuff is stable in next Rust stable.

emilio: I redid bindings for CGL on my spare time,

emilio: let me gist the ffi module.

nox: sure. Not sure if we could differentiate between both.

emilio: gist.github.com/nox/76a6293c4659e12…5ab3eaf2070#file-ffi-rs-L6404-L6428

emilio: We may not need to differentiate between both,

emilio: but I like how this brings us closer to a 1:1 mapping for C.

emilio: I.e. there is no namespacing of enums in C, so we shouldn't namespace them on the Rust side.

emilio: I also did something similar in the public API gist.github.com/nox/defd0d6f60be14c2508337cb57f57e16

emilio: The u32 struct wrapper + associated consts mean we can have extensible enums where adding variants don't break semver.

nox: nice

emilio: Also I had fun with that crate,

emilio: it doesn't depend on anything, not even bitflags.

And in ffi.rs you can see strlen because it also doesn't depend on libstd but core has no CStr.

ajeffrey|uk: What are the new Blob failures?

nox: goes and looks...

ajeffrey|uk: Also I think all the expectations similar to gist.github.com/nox/defd0d6f60be14c2508337cb57f57e16 are obsolete.

Mmmh,

c/p fail.

ajeffrey|uk: github.com/servo/servo/blob/fbbdc51…ansitions/idlharness.html.ini#L2-L3

I think those are from when our SM was so old the stuff wouldn't parse.

And kinda like for WebGL tests from Khronos, this is probably some sort of fail() function that doesn't even get called when the problem disappears, so the expectation update tool can't see that those should be removed.

nox: in general, the test failures are due to promise tests now running.

ajeffrey|uk: Yeah, but that doesn't change what I said I think. :P

nox: yes, I don't know if that applies to the Blob failures.

ajeffrey|uk: tests/wpt/web-platform-tests/css/css-transitions/idlharness.html:2:<title>css-transitions IDL tests</title>

It's the title page, I think a good heuristic would be to double check whether remaining single failures correspond to title pages.

emilio: Also,

emilio: discovered that context switches are not really necessary on macOS.

(OpenGL context switches I mean.)

which claims the blob test failures are real.

ajeffrey|uk: Nah.

ajeffrey|uk: That's exactly what I'm saying.

ajeffrey|uk: This tool only works for stuff that *changed*.

ajeffrey|uk: If you have if (some_test) { fail("that test failed") },

and then you fix a bug that makes some_test be false,

nox: OK, I'll try removing the existing ini files and see if that makes a difference.

the tool won't remove the failure named "that test failed".

ajeffrey|uk: I *think* that you can safely remove any single failure that correspond to the test title.

ajeffrey|uk: Khronos tests are full of this pattern. See commit 6c7609c6f0230c18324ef80fd34caf1bf36ce1c6 where I change no code but remove a bunch of outdated expectations.

nox: oh pox I see what you mean.

For a sec I thought you called me pox,

some may think I'm toxic but come on. :P

nox: oh, how so/

*?

emilio: You on your computer?

nox: yes

nox: removing the existing .ini files then running update-wpt should be sensible?

emilio: Open /System/Library/Frameworks/OpenGL.framework/Headers/gliDispatch.h

ajeffrey|uk: Yeah.

nox: not sure I'd find that on Linux ;)

emilio: Oh. :)

emilio: Let me find them on GH.

emilio: github.com/phracker/MacOSX-SDKs/blo…rk/Versions/A/Headers/gliDispatch.h

emilio: github.com/phracker/MacOSX-SDKs/blo…rsions/A/Headers/CGLMacro.h#L16-L45

emilio: I also googled for a while and find documentation about Quartz Composer plugins where they say to never change the current context and instead use those macros,

and other random people confirming that you can safely call gl functions on inactive contexts with this stuff.

emilio: I think we could experiment with implementing the gleam trait with that stuff if I ever finish the stuff I was doing those last 2 weeks around CGL.

TL;DR macOS has an internal dispatch table with a renderer parameter that we can call ourselves.

ajeffrey|uk: So SM doesn't build with clang? :(

And why is libfontconfig patched?

nox: so that's like a vtable associated to a given context? Makes sense

emilio: Yes.

emilio: Also, I think I know how to implement preserveDrawingBuffer,

emilio: and I think I found the context loss infra in CGL.

nox: I didn't know about context loss in CGL, thought it was mostly an EGL thingie

emilio: It's pretty much the same.

emilio: Not lost / innocent / guilty blacklisted status that gets autocleared on lookup, except for the guilty blacklisted one.

ajeffrey|uk: Could you maybe post an executive summary of the various current hacks in your PR?

There is >300 comments and I don't really want to dig the info after 2 weeks of PTO. :P

lol

nox: SM does build with clang, but getting all our other dependencies to compile with clang took a while.

ajeffrey|uk: So it doesn't need GCC 6 anymore?

nox: the change to libfontconfig is to allow EXPAT_LIB_DIR to be set.

What's the relation between EXPAT_LIB_DIR and libfontconfig and SM?

And could you write all of that stuff in a commit message please?

nox: we use gcc for the x86_64 builds,

Why?

nox: changing as little as possible.

Oh we already use GCC? I'm dumb.

Ok.

We moved to clang where we had to.

ajeffrey|uk: Can you squash the expectations' commits together? Not with the first one but I want to start the review with them.

nox: the annoying one was arm, where we only moved because the gcc x-compilers are stuck on old versions on trusty.

nox: they are all squashed, apart from the ones I'm doing now.

No package 'gstreamer-video-1.0' found

Ugh I can't build Servo anymore.

nox: I'll post an executive summary of the SMup changes.

Manishearth: Why are all those gstreamer plugins needed?

I don't really want to install a package with -bad or -ugly in its name. :/

ajeffrey|uk: Ack thanks.

nox: welcome to gstreamer :)

ajeffrey|uk: But I just want a build.

I don't think the README should require ogg support and whatnot, I doubt Servo doesn't build without it.

Can't find which package provide gstreamer-player :(

nox: Which distribution ?

macOS

Homebrew

Ha

Don't know then but I think there's a PR to help people on mac

Ah, it's the -bad plugins set.

Whatever…

Yeah I saw that but I'm trying to understand what provides what.

AFAICT you don't need the -ugly and the -good ones.

ajeffrey|uk: enumerate/newEnumerate

Reminds me mysql_escape_string and mysql_real_escape_string.

ajeffrey|uk: Good summary, thanks!

ajeffrey|uk: What's JSCLASS_FOREGROUND_FINALIZE out of curiosity?

ajeffrey|uk: Many changes seem to introduce new intermediate variables for no reason, is there an explanation somewhere for those?

ajeffrey|uk: Never mind I can't read. -_-

nox: -good is needed, -ugly well... i think you can avoid it yes :)

philn: Yep it built. :)

-bad is not bad per-se, it's more like a staging area for -good and -base

ajeffrey|uk: `ref mut` bindings mean you don't know at call sites whether the mutability is needed or not.

ajeffrey|uk: Done.

ajeffrey|uk: Will review expectations after lunch.

standups: Reviewed the SMUP PR.

emilio: Just realised something,

emilio: you know how c_void is weirdly defined in libstd and is actually some sort of i8 value right?

emilio: that's because otherwise LLVM gets confused sometimes and misses optimisation opportunities.

emilio: Doesn't that mean the various zero-sized struct types for opaque pointers in bindgen suffer from that issue too?

nox: ta.

nox: JSCLASS_FOREGROUND_FINALIZE is because you need to specify the finalizer style in SM these days, I picked one at random.

ajeffrey|uk: What if I prefer the purple one?

nox: submit a PR :)

nox: I did not, lol. (so, perhaps?)

rust-lang/rust #53521 This might help with script compile times :)

emilio: Maybe. :D I don’t really know. Maybe eddyb knows.

wat

why would I know anything

nox: oh that's easily alleviated, I think?

nox: make them transparent byte wrappers

eddyb: Yes yes, was just wondering.

nox: idk if LLVM still has that bug tbh

eddyb: extern types would work too right?

eddyb: Oh interesting.

nox: probably not right now?

LLVM shouldn't give a rat's... tail... about pointee types

eddyb: Why not right now?

because they're not i8's?

We control what happens with extern types right?

yeah

Ah, why not?

just not on today's nightly

you could open a PR if it helps

Oh ok.

nox: the casts to u32 are caused by jsapi defining the constants to be u8s. searchfox.org/mozilla-central/rev/8…6ce05ce62bc0c882/js/src/jsapi.h#474

ajeffrey|uk: Lol, nice.

That brings me to another topic,

eddyb: do you think it would make sense to forbid or at least warn on casts that change both the size and the signedness of a value?

-1i32 as u64 can be unintuitive depending on what you meant. I think that’s a footgun.

Manishearth: Is there a lint for that in clippy?

jdm: Thanks :)

nox: Addressed most of the smup issues.

ajeffrey|uk: r=moo

nox: left you a present!

nox: excellent!

jdm: Reviewed smup, fought hot garbage on the Twitters, ready for WebGL stuff. 😎

HEY EVERYONE, nox just r+d SMup!

jdm: we still need to update the windows builders?

ajeffrey|uk: oh, you were testing servo master on the update as well. right.

ajeffrey|uk: Squash it, tho :)

ajeffrey|uk: any idea what larsberg did for that update?

jdm: a fresh windows insstall, a fresh msvc install, then just followed the servo build instructions I thinl.

*think

mmm, I see

nox: will do

jdm: is windows3 in the builbot builder bool?

*pool

ajeffrey|uk: no

jdm: OK, I think that's the only remaining blocker.

jdm: unless you want to take a final look-over on the SMup PR.

jdm: OK

we are ready to roll once we've got an updated msvc builder.

:party:

jdm: care to look at servo/libfontconfig #36?

ajeffrey|uk: I have looked at it; could you explain how/why it's used?

jdm: it's in the arm32 and arm64 builds.

jdm: yes that's exactly what I did (just fresh windows EC2 instance)

and then followed the servo windows setup instructions

sorry I can't help this morning - nanny is sick so I'm on kid drop off duty :-(

larsberg: I'm about to start trying to setup buildbot on windows3

sorry to hear that!

jdm: e.g. github.com/asajeffrey/servo/blob/8c…d132/etc/ci/buildbot_steps.yml#L228

jdm: good luck! Follow the instructions as if the fate of that machine depended upon it :-) That said, I've done windows service hacking a lot and can regedit-plunge to fix things up if something goes terribly awry

ajeffrey|uk: we have an expat-sys build; why do we want to ignore it?

otherwise the linker tries to link against the x86_64 expat libs,

huh

jdm: does the expat_sys build include building the lib?

ajeffrey|uk: ah, not always: github.com/servo/libexpat/blob/master/build.rs#L11

we allow forcing a harfbuzz build with an environment variable, but expat doesn't have that

jdm: so we link against the existing libexpat,

which is fine until we do a x-compile.

right, makes sense

ajeffrey|uk: do we use the fontconfig stuff too?

ajeffrey|uk: er, freetype

jdm: no but I was making one env var overridable so I just did all of them while I was there.

ha

jdm: Nice CRASH removal.

nox: now the only failures in that test left seem to be related to the depth test

jdm: "[WebGL test #522: gl.checkFramebufferStatus(gl.FRAMEBUFFER) should be FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT was FRAMEBUFFER_UNSUPPORTED or FRAMEBUFFER_UNSUPPORTED]"

I'm squinting really hard at that,

nox: yeah, I had the same reaction :P

did the test mix "was" and "or" in the message?

Wait no that doesn't make sense either.

tests/wpt/mozilla/tests/webgl/conformance-1.0.3/conformance/renderbuffers/framebuffer-object-attachment.html:437: msg += " or FRAMEBUFFER_UNSUPPORTED";

Oh, ok.

I saw nothing.

standups: Published crates.io/crates/servo-fontconfig-sys/4.0.6

standups: The SMup servo PR is now approved by nox!

jgraham: All of them ! (probably)

jdm: Some failures in your PR.

yep

jdm: oooooooooo

unfortunately I fear I still need to solve servo/saltfs #879

jdm: oh that's annoying, it's blocking deploying new builders?

ajeffrey|uk: it's blocking making any changes to salt

jdm: :(

the buildbot master config doesn't know anything about servo-windows3 yet unfortunately

I guess I could yolo modify the actual config on disk and restart the service

huh, the salt operation definitely took place

jdm: can we change the DNS entry for servo-windows1?

maybe, but let's try my way first

jdm: better if we can avoid DNS hackery

wooo

yesss

:o

what a ridiculous series of steps

standups: set up buildbot on new windows3 builder.

\o/

WindowsError: [Error 2] The system cannot find the file specified

jdm: build.servo.org/builders/windows-msvc-dev/builds/7990 ^

boo, I followed their instructions, and this is how they repay me?

but this means the new builder is working!

(for certain values of "working")

jdm: nice work :-)

And yes, it's incredible how bad the buildbot setups are

hrm did I not globally install rustup? Or possibly it's not on the path in the environment in which the build is being run

hooray, my new house now has internet

jdm: welcome to the late 20th century.

Fake News,

jdm has no house,

only trains.

definitely gonna have to pick up a amazon.com/Power-Wheels-Thomas-Friends-Exclusive/dp/B00F6N06CS for between the kitchen and office to keep the embark/disembarking in effect :-)

lol

only semi-related, I find the thomas the train series actually horrifying and would never recommend to a child

ahaha

larsberg: Can you overclock it?

larsberg: Make it go 7.2krpms.

github.com/cyrus-and/gdb-dashboard looks slick

Hi everyone, I had raised a request on janitor.technology for servo long back but it is still on the waitlist. Can anyone look into it? Thanks in advance. :)

oh, huh. that's very strange.

janx: ^

jdm++

manrajsingh: Hello! Apologies for the long waiting time. We're currently a bit short on resources, so we can't accept as many new users as we'd like. But I'll send you an invite right away because you asked :) please PM me the email address with which you requested the invite

janx: Thank you for looking into it. I've pinged you my email address. :)

jdm: Thank you again :D

hooray, a different windows build failure

manrajsingh: invite sent!

janx: Yup, got it. Thank you so much again :D

no worries :) happy to help

jdm: er, yay I guess

jdm: that failure is caused by not running vcvars64 before running mach.

yes

and I'm trying to sort out why that's the case

given github.com/servo/servo/blob/a4e8164…398041ce4d9c7aed5dea4/mach#L75-L119

jdm: I've always just run it by hand I'm afraid.

jdm: oh I may only have the VS build tools on that machine and not full VS

So that file might not exist on the machine

larsberg: no, it exists

aha

but vswhere isn't returning an installation path

probably for that reason

if we hadn't merged servo/servo c20efd5 then it probably wouldn't affect us

Aha, yeah, it probably is checking a bunch of registry keys that don't get set in that case

bah, vswhere isn't finding _anything_

I'm tempted to revert that commit for now

REVERT, WHAT IN TARNATION

nox: we dont need ugly, we do need bad for gstplayer which we dont use yet but sill

*will

👍

Manishearth: Can't build without -bad, not sure what you mean by not using it.

nox: i dont think theres a lint for that, file it?

Manishearth: SimonSapin mentioned to me there has been discussions for similar topic on the language itself, so I'll file an issue for rust-lang/rust, probably.

nox: its required to build servo media but the player part of servo media is unused by servo

Manishearth: Oh ok.

nox sg

nox: ceyusa is working on video playback stuff in servo

nox: let me know if/when you open that issue

SimonSapin: I'll cc you.

Tempted to name it Heterogeneous Integer Casts Considered Harmful

nox: might be slightly amusing but possibly not productive

SimonSapin: The intent being to put all the snark in the title and make the body be only Positive Mental Attitude.

I think Gankro tweeted something about the title of internals.rust-lang.org/t/pre-rfc-a…-performance-the-raw-entry-api/7043 but I can’t find it now

?

oh right

yeah framing it in a negative way was very non-productive

lots of confusion

i was assuming everyone had the same background details as me

Gankro: Maybe starting the topic with a quote about bitterness didn't help though. :)

(And mentioning morals, I got triggered somewhere else when someone told me a feature was "morally right".)

Anyway, first I need to vanquish my procrastination and file the damn issue, chide me about that not about a hypothetical title. :P

nox: I didn’t mean to chide, just say that like the Church and State, maybe Shitposting and Actual Proposals are better kept separate :)

Heh. :D

SimonSapin: Nah I know, don't worry. I just like the verb "chide" and don't have enough opportunities to use it in my life.

haha

SimonSapin: Probably next time I meet the gang and use it IRL, I'll discover I've been pronouncing it wrongly my entire life.

nox: And they will chide you for it?

jgraham: They will be amused, but not as amused as when I said "gigabyte".

ajeffrey|uk: will you have time to provide feedback on servo/servo #14692#issuecomment-414277683 ?

nox: By the way, I'm in favor of getting rid of the more risky `as` use cases, and eventually deprecating it completely, so I agree with your idea, but...

ALWAYS A BUT 😡

nox: One legit use case that will be made more verbose is assembling a multi-byte value of out `u8` bytes, or going the other way

:P

mbrubeck: Not sure what you mean.

nox: e.g. `(x & 0xFF) as i8 as u8`

I see.

mbrubeck: That truncates though.

Can anything wrong happen on truncatio-- oh yes of course.

Fortunately we're getting more helpers for this in libstd, so most user code shouldn't need to do this "manually"

how to open a new terminal in noVNC in janitor ? I accidentally closed the one present and unable to open now =L

janx: ^

This must be a really silly question, but I have no clue...

nupur: There is no silly questions, only bad answers and confusing user interfaces. :P

nox: hehe this makes it sound better :P

jdm: Sorry been a bit busy, but with #21150 is there any profiling code that can be reused (like `components/profile`)?

good question!

cbrewster: I think you can probably start with time::profile and see how far that gets you - add a new WebGlDomApi ProfilerCategory variant with a sub-enum for each profiled method

jdm: sounds good, thanks!

cbrewster: and then running servo with -p will show the profiling output in the terminal

ha, the windows build failed during packaging

need to install wix or something I guess

jdm: Hi, servo/servo #19797#issuecomment-414143596 I'm not quite sure about how to implement the 'test'. Would a jsfiddle, which I'd run in servo, do the trick?

pylbrecht: a local html file that you load from disk should work fine

jsfiddle and similar sites don't always work in servo yet

jdm: didn't think of that.. thanks

jdm: I'm out at the intern leaving do right now, but I should be able to look tomorrow.

jdm: urg, didn't check packaging.

"what could possibly go wrong?"

:)

jdm: Do you know how the value of message is set here github.com/nupurbaghel/servo/blob/P…script/dom/htmlimageelement.rs#L966

nupur: yes. it comes from the value sent over responder_sender

nupur: which is stored in the list of listeners for a particular cache entry in the image cache

So for a new request, as in our case where we swapped values of iframe

nupur: one thing I didn't think of; the reacting algorithm implementation should probably increase the generation number

nupur: you might be seeing an issue right now where there are conflicting responses arriving

jdm: Yes metadata and none both

Thats what you mean right?

by conflicting

right

jdm: where can we update generation in that case?

nupur: try before calling await_stable_state, like update_the_image_data

okay trying....

jdm: are tests runned under valgrind on CI?

retep007: no

img.bananium.fr/eijebong/7751d9df-8d15-4e28-afd8-cc24672835f2.png Woups, I think I missed something

eijebong: hmm?

a cargo.toml override?

jdm: I forgot some things depending on hyper 0.10

So I just duped everything

oh, ha

lol, tried continuing, duped one more

This is going to be so bad T_T So many update at once

I'm so sorry for the reviewer(s) in advance

wooooo windows try worked

It keeps getting worse T_T

Now openssl is duped

Damsn hasn't the ipc-channel stuff already landed ? :/

jdm: servo/servo #19797#issuecomment-414143596 Unfortunately I couldn't verify my code yet. I looked at RUST_LOG="debug", but there are no entries regarding script thread shutdown. Should I create a PR anyways, so you can have a look at the code?

pylbrecht: link me to the branch?

jdm: github.com/pylbrecht/servo/tree/measure.blocked.layout.queries

pylbrecht: have you tried RUST_LOG=script::script_thread when using a debug build?

jdm: not yet, I'm on it.

I just used RUST_LOG="debug"

05.395567: DEBUG 2018-08-20T19:44:11Z: script::script_thread: tracing fields of ScriptThread

that's the last message

Am I closing servo correctly by clicking the red "x" on the window's title bar? (os x)

huh

jdm: \o/ re MSVC build, let's see if it works with SMup.

pylbrecht: try moving the reporting code to after the script_thread.start() call

jdm: Ok.

emilio: Oh also, I looked into how the hell one is supposed to properly share IOSurface values across processes.

It seems to require a huge infra (XPC), but it could be cleanly done with Rust and some tooling around it.

The only way to have IPC surfaces that are not global (so that other processes can't try to access them) is to have an XPC service, which is a specific kind of macOS bundle, which means servo itself could only be run as an app bundle on macOS afterwards.

hi Patriciap25!

ajeffrey|uk: the try will need to wait until my current PR finishes merging.

I'll run it through once it does.

jdm: it's a spambot

;_;

jdm: Don't you want to meet hot singles in your area though ?

jdm: okay, I can be patient.

standups: investigating weird behaviour in generic script PR.

jdm: placing the reporting code here: github.com/pylbrecht/servo/blob/bdc…onents/script/script_thread.rs#L605 did not work. Debug output is still the same.

pylbrecht: I meant immediately after script_thread.start()

jdm: oh ok, I'll try that then.

pylbrecht: oh wait, that shouldn't make any different

*difference

pylbrecht: ok, what about in the Drop implementation for ScriptThread

jdm: that's a neat idea. Hopefully that will do the trick, because I only have time for one more build before going to bed!

yeah, I know that feeling

So, I have a question, how am I supposed to impl MallocSizeOf on that github.com/hyperium/mime/blob/master/src/lib.rs#L52 ?

Everything is private

Is it mime.type_().len() + mim.subtype().len() + 1 (the /) + 1 (the usize) + 2 (Option<usize>) + sizeof(params) whatever that is ?

eijebong: #[ignore_malloc_size_of]?

jdm: But I don't want to break everything :/

eijebong: break in what sense?

jdm: I dunno, I have no idea what mallocSizeOf is used for

But I don't want to update stuff and remove features

ah, I see we have an implementation in-tree right now

Yeah

eijebong: stuff like the declaration you linked to where it's private is too risky to try to guess like that

eijebong: because stuff like github.com/hyperium/mime/blob/master/src/lib.rs#L79 doesn't consume any memory

jdm: Yeah I just saw it

eijebong: adding ignore_malloc_size_of will just mean that our memory profiling tools will be a bit less accurate

jdm: Ok, so just remove it then and add #[ignore_] where needed

yep

I hope I won't have too many of those

Nice ^

jdm: still the same output..

I'll comment the issue including my .html (maybe that is wrong all along?)

would you mind commenting on the issue if you have any new ideas, because I'm going to bed now.

smup failed again, but not on windows

gotta update the test expectations before we sync from upstream tests again

bah, why is servo-mac9 taking almost 50 minutes to build servo

that's an absurd number

Is it ?

newbie question - where is cascade() called in servo? both as in, where in the source, and on which thread (a script thread? some other thread?)

cascade() being the main function exported from properties.mako.rs, not CustomPropertiesBuilder::cascade()

cdisselkoen: In the source, the call is here github.com/servo/servo/blob/master/components/style/stylist.rs#L867

mbrubeck: ah yes, thanks, that's what I was looking for

the turbofish spoiled my grep

cdisselkoen: In terms of threading, it's done on the layout thread pool, I think here github.com/servo/servo/blob/master/…ts/layout_thread/lib.rs#L1307-L1319

cdisselkoen: This page has a high-level overview of the tree traversals that the layout threads perform: github.com/servo/servo/wiki/Layout-Overview

Yup, its called by style_resolver.rs i think

(Cascading is part of the first traversal, "style computation")

thanks!

Its called here github.com/servo/servo/blob/master/…onents/style/style_resolver.rs#L306

@mbrubeck is that Layout overview still up to date?

JaseW: Not 100%, but I think only minor-ish details have changed.

1 test fail on mac ?!

so close

jdm: fingers crossed

Question: Hyper removed the RawStatus thingy and now only has StatusCode (which can't have text). Should I just remove all mentions of status reasons ?

(Or do I transform everything into (StatusCode, String)

remind me what we use status reasons for?

No idea if they're even used in the end

I know they're passed around everywhere

eijebong: it looks like the only usage we might care about is in the devtools

eijebong: so it would be good to keep it, I think

jdm: Oukay'

Will do

But now it's probably time to take a break and go to bed

crowbot1: what does the web need?

fair

indeed

holy shit

you heard it from me first

:3

OMG!

and jdm isn't even here

ajeffrey|uk++

larsberg++

jdm++

crowbot1: tell jdm jdm++

crowbot1 is not happy

crowbot1: botsnack

OMG! Great work ajeffrey|uk, jdm, nox, etc. :-)

I DID NOTHING BUT A TWITTER-RAGE-FUELLED REVIEW FAM

il pushait du smup debout

Ah ah.

nox: nichijou gif clearly contributed

Ah ah ah ah.

would future smups be easier at this point?

AFAIK yes, because ajeffrey|uk doesn't give fish, he gives fishing lessons.

ajeffrey|uk++

ajeffrey|uk += 1 million

ajeffrey|uk is gonna wake up tomorrow with a million notifications and go on pto a happy man

Oh he is on PTO next week?

That's nice.

wednesday IIRC

Cool.

quite timely

definitely earned it

we clearly need an ajeffrey gif involving dr. who, a giant monkey covered in spiderwebs getting bonked, and something implying vacations

larsberg: ah, too bad jdm is gone, otherwise it'd be done in servomemes already

I'm obviously too lazy to make memes.

unless forced

HAHAHA THAT WAS A BAD PL JOKE ok I'll show myself out

Ah ah ah ah.

hehe

:D

too strict around here anyways

mbrubeck: You wanted to make PL laziness jokes. ^

everybody always judging and evaluating me

sounds exactly like what SM does

that's just the type of community we are, I guess