bholley: Manishearth: What about something like ^?

emilio: hmmmmm

emilio: sounds like it could be fragile, but if it works ... !

I'm wary of this being annoying to contributors but idk

(as opposed to just pushing on "rustfmt everything")

Manishearth: well, the problem with pushing on "rustfmt everything" is that as soon as other people land code without using rustfmt it's useless

Manishearth: probably we shouldn't land that as-is though, probably should at least warn on test-tidy or what not

Manishearth: which is blocked on the `rustfmt` "check" stuff.

emilio: no I mean that we CI-check rustfmt everything

Manishearth: that basically means that whenever rustfmt changes anything CI is broken

Manishearth: which is not acceptable

emilio: use the pinned-nightly rustfmt

Manishearth: (IMO, at least :P)

./mach fmt

Manishearth: sure, but then you need to periodically update it

*shrug*

emilio: either way, please provide a mach command so that folks can easily run this step locally

Manishearth: sure, sure, that's not intended to be mergeable as-is

sgmt

nrc: fyi #20617 is my proposal for doing rustfmt CI on servo

emilio: these

ctedJob=172996817 are fixed by be94036426d7?

how do I run one unit test with mach? I want to run components/net/tests/cookie.rs

./mach -v cargo test does it

./mach cargo test-unit works. I was trying with ./mach cargo test unit

hmm... pointer cannot be type of global static?

oh, it can. I mean, type of const

SimonSapin: is incremental compilation a lot faster? Just realized how much space it takes on my disk. Not sure if it's worth it or not.

why "static XXX: *const Type" doesn't work but "static mut XXX: *const Type" works...?

build.servo.org/builders/mac-dev-un…ilds/11642/steps/compile/logs/stdio Hu ? :/

xidorn: What do you mean?

nox: you can see #rust channel for more

Paul: if it's not you should talk to mw on #rustc :P

emilio: Somehow I don't see Josh's bot do the WPT sync PRs anymore -.-

I hope I didn't break the hell out of it.

nox: Speaking of breaking bots, have you seen my PR ? :p

eijebong: Which one?

nox: servo/servo #20611

Paul: "a lot" really depends, try disabling it with `CARGO_INCREMENTAL=0` and see

IMO space on a disk paid by mozilla is much cheaper than my time

nox, review please ^ :)

marmistrz: Ack, will review soon.

hello

wanted to help with rust-bindgen

which issue should i pick up?

emilio: I wonder if you should have created a FFI WG for bindgen,

you get your workgroup, she gets her workgroup, everyone gets their workgroup.

nox: lol

nox: I want a dependency updating workgroup

nox: just became editor of the CSSOM spec, don't ask

:O

emilio: NICE

emilio: So you can't hack on Mosaic anymore? :(

nox: looks like it

emilio, you actually hack on Mosaic?

qweo: lol nope

qweo: nox is just french :)

Ah ah.

It runs on modern systems, so I thought maybe it required porting and whatnot

Joke is lost on me though

qweo: For once unrelated to my Frenchness, to be fair.

qweo: It's just that emilio happens to also have some commits in Webkit and Blink,

qweo: so I once joked he also has push rights to Edge, and Opera, and Mosaic,

and even Tim Berners-Lee's genome.

Was WWW the original browser ever open-sourced by the way?

Which engine is easier to hack on? You'd likely say "Servo", so the next-best is interesting too

emilio: ^

qweo: Servo is the only one in Rust, all the others are in C++, so all the others are significantly harder to contribute to, IMO.

blink is stupidly massive

if you like wrappers, you'll like it

Significantly more so than WebKit and Gecko?

qweo: Yeah, blink's build-time is terrible.

qweo: I'd say Servo -> Gecko -> WebKit -> Blink

qweo: however I like WebKit's code style much better than Gecko's and Blink's

Which aspects of it? Not mere formatting I presume?

Not being >20 years old help.

nox: well, sure. Blink is a fork though, and I still like it more

emilio: True.

qweo: mostly conventions... the code style generally feels cleaner, I dislike gecko's aFoo for arguments and such, and blinks combination of snake_case and CamelCase

why_is_it_called_snake_case_by_the_way?

Because ____ is a snake on the ground.

snake🐍case

Is there a crossed-out camel emoji? :-)

Twitter "UX" when openned in a text browser is attrocious

qweo: There is no such thing as text browser.

Text-mode. And there is, if you want to call pager a text browser :-)

nox: What about baguette🥖case ?

eijebong: It's not an established casing scheme.

because kebab case is one . o_o

?

Yes it is.

It's-with-dashes.

Pretty popular in Lisps.

Oh ok...

eijebong: Think the meat on the spinning thingie.

"meat"

It's meat.

emilio: "If first is negative, an INVALID_VALUE error will be generated."

emilio: next method: "in addition the offset must be non-negative or an INVALID_VALUE error will be generated"

Y U NO WRITE IT THE SAME WAY

nox: lol

xidorn: I'm cssom editor now lol

emilio: oh? wg just resolved something there?

xidorn: yup!

emilio: cong :)

xidorn: well, CSSOM seems like a can of worms

emilio: you should push a commit to add yourself into the list first

xidorn: but hopefully now I can do something about it

xidorn: fair enough :)

emilio: twitter.com/nokusu/status/984071863667167234

nox: lol @ hair model

emilio: Speaking the truth.

emilio: Maybe that's why you work on hairy code.

nox: worse so far, and the bar was already low :D

emilio: I'm a limbo dance god.

hah

ajeffrey: r? #20579

bholley, emilio, SimonSapin: rust-lang/rust #49879

(Issue, not PR, unfortunately. :P)

nox: \o/

nox: k17e mentioned yesterday he was going to have mw look into it

k17e?

Oh, I think I know who you mean.

Or not, confused myself with a Japanese Erlang dev.

kentuckyfriedtakahe

Anthony Jones

Hah.

(mw's manager)

bholley: Ack.

avadacatavra: are you the right person to ask about anything to do with metrics?

I'm failing to work out this test failure: servo/servo #20579#issuecomment-380501491

avadacatavra: in particular, I don't understand github.com/servo/servo/blob/8281ee0…components/metrics/lib.rs#L339-L371

it looks like if the first-paint is received before the first-contentful-paint then two first-paint events will be generated (which would explain the test failure).

cbrewster: ^

ajeffrey: the paint time metrics were written by ferjm iirc but he's out

so i would double check spec and then i can r the pr

avadacatavra: OK

just because i'm not familiar with the paint time part at all and seems like you might have already figured out the problem :)

Hmm, github.com/servo/servo/blob/8281ee0…5610c6bde/components/metrics/lib.rs has no spec links.

bholley: the plan is to figure out exactly what needs to happen, then figure out how to make it happen

inlining between C++/Rust will require clang which is currently 5-10% slower than msvc

but hopefully the same inlining will reduce the gap between msvc and clang

standups: trying to track down test failure in performance metrics, which is blocking session history cleanup. servo/servo #20579#issuecomment-380501491

cbrewster: #20579 landed \o/

ajeffrey: \o/

standups: Filed issue #20625: Intermittent fail in /_mozilla/mozilla/paint_timing.html

cbrewster: yay, about:blank is a bit less horrible!

ajeffrey: I’ve been wanting that fix for quite a while now

cbrewster: me too!

standups: Created github repo for work on speculative evaluation github.com/chicago-relaxed-memory/spec-eval

ajeffrey: how does this look for paint metrics: docs.google.com/document/d/1BR94tJd…poeet0XoTW60M0SjvOJQttKT-JK8HI/view

avadacatavra: do we need stuff that isn't in w3c.github.io/performance-timeline ?

afaik time to interactive and paint metrics are part of progressive web metrics

which i'm not sure has been standardized yet

so w3c.github.io/paint-timing ?

avadacatavra: ^

ajeffrey: ah yeah that looks more official :p

Oh btw, I'll need help with #20611 :/ That's a loooooot of failing tests

avadacatavra: and the rather less official github.com/WICG/time-to-interactive

ajeffrey: i still don't see a better link for tti

ajeffrey: agreed

pr incoming shortly :)

standups: Attended meeting on personal agents (with my angle being MR security).

ajeffrey: "personal agents"?

pardon me but that sounds like something that's taken directly out of a 007 movie

KiChjang: think Cortana

ohhh

KiChjang: with my security hat on, I'm thinking "trust anchor".

we are slowly pacing towards the future that megaman battle network paints

so if your immersive reality says "honest I'm your bank", you have less reason to believe it than if your personal agent introduces you.

where users navigate the virtual online world with a NetNavi

ohh, interesting stuff

so it can play both roles i believe?

both authentication of the user

and the website being visited