02:12
<MikeSmith>
foolip: I see that Denis responded to some (but not all) of your review comments at https://critic.hoppipolla.co.uk/r/604
02:13
<MikeSmith>
foolip: if you could check and see if any of the remaining open issues in that review can be close that would be great
02:13
<MikeSmith>
foolip: that way we can get them down to just the ones that are actually still waiting on Denis
02:14
<MikeSmith>
foolip: and then I can ping Denis and ask him if he can make time to look back at them
02:15
<MikeSmith>
foolip: we're no longer paying Denis to work on the testsuite, and this is the last PR we still have open from the time when we did have him working on tests
02:34
<MikeSmith>
so the "is" attribute is not actually formally defined anywhere yet, right?
02:35
<MikeSmith>
at least not in http://w3c.github.io/webcomponents/spec/custom/
02:35
<MikeSmith>
it's mentioned there but never defined
02:35
<MikeSmith>
dglazkov: ↑☃
03:12
<caitp>
hixie linked me to a sufficiently normative definition of it, but I can't recall where it was, @MikeSmith
03:13
<caitp>
there are like dozens of those specs, makes it really hard to find things =)
03:14
<MikeSmith>
caitp: yeah
03:14
<MikeSmith>
caitp: OK thanks I guess I'll give up looking for it for now
03:14
<caitp>
if i still have the bugmail i can probably dig it up
03:15
<caitp>
mm, nope, gone
04:34
<TabAtkins>
Hixie: We just agreed to kill the image-orientation property, so we need the exif autorotate attr. (dbaron needs the functionality *somehow*, and wants to make sure it's a sure thing).
04:34
<TabAtkins>
http://lists.w3.org/Archives/Public/www-style/2013Jul/0568.html
04:35
<TabAtkins>
wrong window, sorry
05:42
<mikey85>
hello everyone :)
05:42
<mikey85>
God bless everyone :)
05:42
<mikey85>
message me if you would like to make friends with a good hearted Christian :)
05:45
<MikeSmith>
TabAtkins: wrong window again man, plus you've now exposed you super-secret mikey85 alternate nick
06:31
<hober>
MikeSmith||
06:31
<hober>
err, ++ even
06:40
<sangwhan>
MikeSmith: smooth
07:04
<TabAtkins>
MikeSmith: That's hardly even a secret.
07:06
<zcorpan>
Hixie: i notice there's an "HTML - <img>" component in bugzilla now. maybe the bug filer should select that component for img bugs? and i guess i can move the existing ones
07:07
<MikeSmith>
zcorpan: yeah Hixie asked me to make that component for exactly what you just described
07:10
<zcorpan>
ok cool
07:12
<zcorpan>
Hixie: for the record i don't like the fading thing. i'd prefer if the styles are stable so i don't have to keep moving my mouse while reading and not lose track of where the links are
07:12
<zcorpan>
Hixie: i'd be OK with having the links always be black and underlined, or some such
07:15
<TabAtkins>
Fading thing?
07:15
<TabAtkins>
Oh, I saw it.
07:18
<TabAtkins>
But yeah, agree that the fading thing is annoying. At least keep the underlines.
07:21
<zcorpan>
i'd find the fading annoying even if it keeps the underlines. it's distracting me that it fades back and forth
07:34
<Ms2ger>
gsnedders, IE6 still supported? I missed it
07:53
<MikeSmith>
I like the fading thing. But I guess I'm weird.
07:54
<MikeSmith>
But I like the colors too, so I'd rather have the colors than having the links always be black or whatever other compromise
07:55
<MikeSmith>
and the complainers that are color-averse would just have to find some way to live with it
08:03
<annevk_>
http://www.openbsd.org/papers/bsdcan14-libressl/mgp00013.html whoa, OpenSSL had EBCDIC support
08:42
<sangwhan>
wonder if libressl is nuking that perl generated assembly nonsense in openssl
09:01
<MikeSmith>
can the empty string be a valid "source size list"? http://picture.responsiveimages.org/#valid-source-size-list
09:01
<MikeSmith>
is that BNF? does the lack of brackets around something mean it's required?
09:01
<MikeSmith>
what does "#" means?
09:14
MikeSmith
finds "A hash mark (#) indicates that the preceding type, word, or group occurs one or more times, separated by comma tokens (which may optionally be surrounded by white space and/or comments)." http://drafts.csswg.org/css-values/#component-multipliers
09:14
<MikeSmith>
good times
09:15
<mounir>
annevk: I believe the array thing is on purpose
09:16
<annevk>
mounir: can you restate that?
09:19
<mounir>
annevk: regarding navigator.languages
09:19
<mounir>
it returns an Array that has to be the same unless the values have changed
09:19
<mounir>
so you get previousValue == navigator.languages
09:30
<annevk>
mounir: so you return a JavaScript Array?
09:30
<annevk>
mounir: the IDL says otherwise
09:30
<mounir>
annevk: that's a Gecko quirks
09:30
<annevk>
mounir: in the spec
09:31
<mounir>
annevk: the spec says DOMString[]
09:31
<mounir>
annevk: this is what my Blink patch does
09:31
<mounir>
annevk: and in Gecko, sicking said that for some reasons, the right way was sequence<DOMString>
09:31
<mounir>
because of binding generator or something
09:32
<Ms2ger>
mounir, DOMString[] is bad
09:33
<mounir>
Ms2ger: don't be judgmental with that poor thing
09:34
<tobie>
Is there a palatable doc somewhere that explains the diff between DOMString[], sequence<DOMString>, Foo extends Array and the like?
09:34
<annevk>
mounir: [] is an IDL bug and needs to die
09:34
<mounir>
annevk: I understand that
09:34
<annevk>
tobie: not really, mostly we need someone to maintain IDL
09:34
<Ms2ger>
Then why are you using it in blink?
09:35
<tobie>
annevk: the language in WebIDL is NOT palatable.
09:35
<Ms2ger>
tobie, there will be no need to explain DOMString[] if DOMString[] is gone, though
09:36
<tobie>
Ms2ger: right, but in the meantime would be useful.
09:36
<tobie>
Ms2ger: I expect it's going to last a while before it is gone.
09:39
<annevk>
tobie: depends on when IDL becomes maintained again
09:40
<tobie>
annevk: sure, but we're still going to see references to it all over the place.
09:41
<annevk>
tobie: except at that point it'll be "oh this is invalid IDL, please fix"
09:41
<Ms2ger>
We can't, it's in CR
09:41
<tobie>
^ that.
09:41
<annevk>
lol
09:42
<Ms2ger>
We'll just reference a WebIDL draft from 2008
09:42
<tobie>
or already published as a REC, etc.
09:42
<annevk>
something something http://annevankesteren.nl/2012/11/process
09:45
<tobie>
I thought heycam was maintaining WebIDL (and that he currently was on holidays, hence the delay)
09:46
<Ms2ger>
He's not exactly quick to respond when he's not on holiday either
09:47
<annevk>
heycam has a ton of other responsibilities
09:47
<annevk>
IDL at this point requires about three to six months FTE
09:55
<annevk>
http://www.w3.org/TR/DOM-Level-3-Events/#widl-DocumentEvent-createEvent o_O
09:55
<SimonSapin>
sangwhan: http://youtu.be/GnBbhXBDmwU?t=51m9s
09:59
<darobin>
heh, annevk talking in FTEs
09:59
<darobin>
you can almost feel the manager waking up in there :)
09:59
darobin
ducks
10:08
<annevk>
mounir: I guess I still don't understand; are we returning a JavaScript Array?
10:08
<annevk>
mounir: seems like in Gecko we are; what about Blink?
10:10
<annevk>
https://www.w3.org/Bugs/Public/show_bug.cgi?id=25800 #w3confusion
10:11
<mounir>
annevk: isn't what the spec says?
10:11
<annevk>
mounir: no
10:15
<mounir>
annevk: the spec says DOMString[], how isn't that a JS array?
10:15
<mounir>
I'm confused
10:15
<annevk>
mounir: it's an IDL array
10:15
<mounir>
annevk: and that's not seen as an array in JS?
10:15
<annevk>
mounir: correct
10:15
<annevk>
mounir: read http://heycam.github.io/webidl/#es-array and weep
10:16
<mounir>
oh boy...
10:19
<mounir>
annevk: btw, the sequence<T> of Gecko is a hack in the sense that WebIDL doesn't allow returning sequence<T> from an attribute
10:20
<annevk>
yeah I know about that; I don't like how IDL has the same "types" for arguments and return values
10:22
<sangwhan>
SimonSapin: oh boy, the perl is still there
10:24
jgraham
is quite confused about why you would want different types for arguments and return values
10:24
<annevk>
jgraham: the argument one is not a type, it's a coercion
10:25
<jgraham>
A coercion to a type
10:25
<mounir>
annevk: seriously, this "platform array" thing is pretty confusing
10:26
<annevk>
mounir: you can consider it dead
10:26
Ms2ger
promotes mounir to WebIDL editor
10:26
<mounir>
Ms2ger: if you do that, don't complain when you become a Gecko DOM peer ;)
10:27
<Ms2ger>
Ha
10:27
<Ms2ger>
Not like I have time to do Gecko reviews anyway
10:28
<mounir>
Ms2ger: you wouldn't be the first reviewer without time for reviews ;)
10:29
<Ms2ger>
Hmm
10:29
<Ms2ger>
Should window.HTMLAllCollection exist?
10:29
<Ms2ger>
I guess so
10:34
<mounir>
annevk: what's the simplest/quickest way to check if an array is a js-array?
10:35
<annevk>
mounir: use isArray maybe
10:36
<mounir>
annevk: that wouldn't return true for a non-js array?
10:36
<annevk>
mounir: it shouldn't
10:36
<mounir>
it seems that the Blink bindings return a JS array then
10:37
<annevk>
mounir: another test you could do is push a number or string on the array and see if it's there
10:39
<mounir>
annevk: working in Blink, failing in Gecko
10:39
<annevk>
weird
10:40
<mounir>
annevk: it says that l.push() isn't extensible in Gecko
10:41
<annevk>
mounir: but Array.isArray(obj) returns true in Gecko?
10:41
<mounir>
yep
10:41
<annevk>
o_O
10:42
<annevk>
"ask bz"
10:46
<mounir>
annevk: I probably did something wrong ;)
11:28
<foolip>
MikeSmith: I've commented on the open issues
11:37
<tobie>
Sounds like I'm not the only one utterly confused by the []/sequence, etc. mess.
11:37
<MikeSmith>
foolip: thanks much
11:39
<tobie>
annevk: isn't sequence<foo> just saying: this function/attribute will duck-type?
11:39
<TabAtkins>
So, if I'm returning a Promise which will resolve to a JS array of FontFace objects, the right type for the return value is Promise<sequence<FontFace>>, right?
11:40
TabAtkins
put that in Font Loading today, so he hopes it's right.
11:40
<tobie>
yup
11:40
<TabAtkins>
kk
11:41
<tobie>
actually, I'm not sure you can use sequence<foo> for return types.
11:42
<annevk>
tobie: as argument it means the implementation will iterate over the object (using Symbol.iterator) to get the list of values; as return value it means JS Array
11:42
<TabAtkins>
ARGH
11:42
<TabAtkins>
annevk: Okay, cool.
11:42
<tobie>
oh. OK.
11:45
<tobie>
It would be helpful to use different syntax for these two different things.
11:46
<tobie>
Why don't we use Array<Foo> for the return type?
11:57
<tobie>
Looking at the WebIDL bug tracker, I get a better sense of what annevk, Ms2ger were referring to.
12:02
<barnabywalters>
anyone here know if work to add native Promise support to XMLHttpRequest is being done somewhere?
12:03
<zcorpan>
barnabywalters: i think that will be called "fetch()"
12:03
<annevk>
tobie: see above where I said we should have different syntax for argument / return value (or get vs set)
12:04
<barnabywalters>
zcorpan: interesting, do you have a URL to a spec? I couldn’t find one easily, might not be looking in the right places
12:04
<zcorpan>
barnabywalters: i saw something in ServiceWorker but it's wasn't fleshed out when i looked at it
12:04
<gsnedders>
Ms2ger: Yeah. I mean, pretty much everyone using IE6 isn't using a server OS, but still. :)
12:05
<tobie>
annevk: yeah. I guess it's inevitable with a not strongly typed language.
12:05
<annevk>
tobie: ToPromise / Promise; Iterable / Array
12:06
<tobie>
annevk: on second thought, I'm not even sure strong type has anything to do with it.
12:06
<tobie>
annevk: ++
12:07
<tobie>
annevk: how does that work for attributes which are rw?
12:07
<annevk>
tobie: I think for properties we need to be more clear on whether they have getter/setter or are a data property
12:07
<annevk>
tobie: currently everything is a getter/setter, but that seems somewhat wrong
12:09
<barnabywalters>
zcorpan: hrm okay, might try diving into ServiceWorker then
12:09
<tobie>
barnabywalters: there's also the fetch spec.
12:09
<barnabywalters>
tobie: that sounds good — link?
12:09
<annevk>
barnabywalters: the idea is fetch(Request req).then(Response r => ...)
12:09
<barnabywalters>
http://fetch.spec.whatwg.org/?
12:10
<tobie>
barnabywalters: yes
12:10
<tobie>
but the fetch JS API isn't defined there.
12:10
<annevk>
barnabywalters: that's where the API will eventually end up, currently it's just defining the network stack
12:10
<barnabywalters>
annevk: ah so xhr is being split into Request and Response objects now?
12:10
<annevk>
barnabywalters: roughly
12:11
<annevk>
barnabywalters: if you look at http://xhr.spec.whatwg.org/ you'll see it's defined in terms of Fetch too
12:12
<TabAtkins>
annevk: I'd be happy if WebIDL arguments were all named interface-like and return types were named object-like.
12:12
<barnabywalters>
annevk: tobie: thanks, I’ll have a read through those
12:13
<annevk>
TabAtkins: I'm not sure what that means I'm afraid; we do have some ideas on revamping the whole interface / [NoInterfaceObject] mess in terms of classes and such
12:38
<zcorpan>
MikeSmith: pointer? https://critic.hoppipolla.co.uk/showcomment?chain=3809
12:41
<zcorpan>
MikeSmith: i can't see anything in www-archive :-(
12:46
<TabAtkins>
annevk: Interface-like means names like "iterable" and "promise-like" and whatnot - adjectives, mostly. "object-like" are nouns.
12:46
<TabAtkins>
Like difference between "functor" (terrible name, also a noun, so misleading) and "mappable" (accurately describes, indicates that it's a quality of an object)
12:50
<zewt>
functor: callable
12:50
<TabAtkins>
That's part of why the name is so terrible, because it has nothing to do with "function".
12:51
<zewt>
i guess functor is specifically a callable object (in C++), so in that case being a noun is correct
12:51
<TabAtkins>
(A function *is a* functor, but for reasons that have nothing to do with calling.)
12:52
<TabAtkins>
A functor is any object with a .map() operation or equivalent - it holds some values/values in a container or context, and lets you pass functions inside of it to operate on the inner values and return the same context with the return values.
12:52
<zewt>
i'm only talking about c++; i've never seen "functor" used in any other context (and I don't know why anyone would)
12:53
<zewt>
though I guess people call c++ functions functors too, for templates
12:53
<jgraham>
functor in C++ means something entirely different to Haskell I think
12:53
<TabAtkins>
Functor is a category theory term.
12:53
<jgraham>
(I think that mappable is a terrible name)
12:53
<TabAtkins>
It's a thing you can map. What better name could it have?
12:54
<jgraham>
Well the problem might be that map is a bad name
12:54
<zewt>
without knowing anything about it in advance, it sounds like "can be used as a the key in a dictionary"
12:54
<zcorpan>
TabAtkins: did the @charset use counter ever materialize?
12:54
<zewt>
(the effect of hashable in python)
12:56
<TabAtkins>
zcorpan: No, I haven't actually coded any Blink since the fork.
12:57
<zcorpan>
TabAtkins: ok
12:57
<annevk>
thanks tobie for the ScalarValueString patch
12:57
<TabAtkins>
zewt: The word "map" is pretty common in functional languages (including JS) to mean "call this on every element of the {array, set, etc} and give me a new container with the results".
12:57
<TabAtkins>
Python's map(), JS's Array#map, etc.
12:57
<TabAtkins>
JS also has a Map, to be sure, but Python calls it Dict to avoid confusion.
12:58
<TabAtkins>
(Also, a Map is a functor. ^_^)
12:58
<zewt>
TabAtkins: a terrible mechanism (python's comprehensions are 100x more readable), but i've never seen "map" adjectived
13:00
<zewt>
okay, i only just noticed the "Show: [x] inherited" checkbox on the right of these docs http://msdn.microsoft.com/en-us/library/system.windows.controls.treeviewitem(v=vs.110).aspx
13:00
<TabAtkins>
Comprehensions are great, but not a replacement in all cases.
13:00
<TabAtkins>
[foo(bar) for bar in bars] is worse than bars.map(foo)
13:00
<zewt>
been growling at these horrible docs for making me squint through endless pages of inherited mess for days because they decided to have an obscure "[x] show pages of useless crap" default tucked off to the side
13:01
<zewt>
TabAtkins: i prefer it, to me the extra typing is worth not having to parse two different patterns
13:02
<zewt>
it's also nicer to have (foo(bar) for bar in bars) and {bar: foo(bar) for bar in bars} without having to change modes when i want a different type
13:03
<zcorpan>
all i hear is "foo bar foo bar foo bar"
13:03
<TabAtkins>
Yeah, the other comprehensions are quite nice.
13:06
<zewt>
incidentally, special thanks to c# for adding comprehensions... in a different order, from foo in bar select foo
13:06
<zewt>
just to make sure it's a pain for everyone to context switch
13:07
<darobin>
I always thought "comprehensions" was what you wish you had when you saw Python code
13:07
<jgraham>
darobin: Nah, it's what ruby doesn't give you
13:08
<zewt>
darobintroooooll
13:27
Ms2ger
wonders if http://code.google.com/p/chromium/issues/detail?id=43394 is ever going to land
13:31
<zcorpan>
Ms2ger: "any time now"
13:56
<TabAtkins>
Note that it's still receiving useful activity. We keep poking at it and trying it out, but it keeps causing perf regressions.
13:56
<TabAtkins>
But they're getting smaller and smaller.
13:59
<zcorpan>
MikeSmith: w3c-test:mirror seems like it doesn't bite for me again :-(
14:00
<Ms2ger>
Without a space?
14:00
<zcorpan>
MikeSmith: https://github.com/w3c/web-platform-tests/pull/996
14:00
<zcorpan>
should there be a space?
14:00
<Ms2ger>
That's what I've done, I think
14:01
<zcorpan>
no dice
14:01
<jgraham>
zcorpan: There is (theoretically) never any point in putting w3c-test:mirror on your own PR
14:01
<jgraham>
If you have permissions to mirror stuff your own PRs should be automatically mirrored
14:02
<jgraham>
So you either don't have permissions or (more likely) the script is broken
14:02
<zcorpan>
yeah ok
14:04
<mathiasbynens>
hsivonen: validator.nu seems down again
14:07
<annevk>
JakeA: we need a story for fetch() outside service workers
14:08
<annevk>
JakeA: I don't think we can squat a global name like that
14:08
<annevk>
JakeA: is it going to be navigator.fetch() outside of workers?
14:10
<JakeA>
annevk: why is window.fetch bad? Likely to be taken by app code?
14:11
<annevk>
JakeA: yeah, taking global names is icky
14:11
<annevk>
JakeA: I'm open to try it I guess
14:11
<JakeA>
annevk: What about Cache and caches?
14:15
<zcorpan>
fetch seems a bit weird to put on navigator
14:15
<annevk>
JakeA: dunno
14:15
<Ms2ger>
location.fetch()
14:15
<jgraham>
Ms2ger is winning
14:15
<zcorpan>
URL.fetch() ?
14:16
<jgraham>
zcorpan takes the lead
14:16
<Ms2ger>
Blob.fetch()
14:16
<zcorpan>
Fetch.fetch()
14:16
<jgraham>
Ms2ger crashes into a wall
14:16
<Ms2ger>
:D
14:16
<JakeA>
f.etch()
14:16
<Ms2ger>
.sketch()?
14:17
<zcorpan>
JakeA: f is unlikely to be taken, seems ok
14:17
<jgraham>
This has gone very Wacky Races all of a sudden
14:17
<jgraham>
With JakeA playing Dastardly
14:18
<JakeA>
haha
14:19
<JakeA>
window.fetch, caches, Cache would obviously be my first choice…
14:20
<JakeA>
window.fetchXML()
14:20
<JakeA>
but it can be used (and will mainly be used) to fetch not-XML
14:20
<gsnedders>
jgraham: Do you want me to rewrite the history of https://critic.hoppipolla.co.uk/r/287 before you review it?
14:20
<Ms2ger>
XMLHttpRequest.fetchHTML()
14:20
<JakeA>
haha
14:21
<Ms2ger>
XMLHttpRequest.fetchHTMLSpdy()
14:21
<JakeA>
The problem with putting it on navigator is it won't be on navigator in the SW
14:21
<jgraham>
The problem with putting it on navigator is that it makes no sense
14:22
<jgraham>
It just ends up as a duming ground for things we were too scared to put on window
14:22
<jgraham>
*dumping
14:22
<jgraham>
gsnedders: Yes
14:22
<JakeA>
like navigator.serviceWorker :D
14:23
<jgraham>
Yeah, pretty much. Things only make sense on navigator if they don't depend on the Window
14:30
<annevk>
jgraham: navigator is a dumping ground
14:38
<jgraham>
annevk: Perhaps, but it's not good
14:43
<Domenic>
annevk: you could block on modules!??!
14:43
<annevk>
Domenic: block service workers on modules?
14:43
<Domenic>
annevk: I assume fetch-outside-workers doesn't block service workers...
14:47
<jgraham>
You could add a .send() method to Request
14:49
<annevk>
new Request(...).send().then(r => ...)
14:50
<Domenic>
hmm
14:50
<gsnedders>
jgraham: there's no prepare rebase link on Critic on that page?
14:50
<tobie>
Request.fetch?
14:50
<Domenic>
feels enough like XHR to trigger some bad feels
14:51
<Domenic>
Request.fetch seems better
14:53
<tobie>
Would be nice to ship fetch outside of SW at the same time (or even before) it is shipped within SW.
14:53
<jgraham>
Do you "fetch" a Request? Surely you "send" a Request?
14:53
<jgraham>
gsnedders: No, you do the opposite
14:54
<jgraham>
gsnedders: https://github.com/mozilla/servo/wiki/Github-&-Critic-PR-handling-101
14:54
<gsnedders>
jgraham: huh?
14:54
<Domenic>
if it's a method on Request.prototype, then send. But if it's a static factory method, then fetch seems better.
14:54
<tobie>
^ that
14:54
<annevk>
you fetch a response using a request
14:54
<tobie>
IO.fetch
14:54
<jgraham>
Request.fetch(request) is going to sound very odd
14:55
<annevk>
jgraham: agreed
14:55
<gsnedders>
jgraham: the first commit on the branch and not the merge-base?
14:55
<jgraham>
Even though it is probably no more verbose than other options, it feels like it is due to the repetition
14:55
<jgraham>
gsnedders: Do your rebase, force push, follow the steps to update critic.
14:56
<gsnedders>
jgraham: I'm following the steps. "Ask for help if this step is daunting"
14:56
<jgraham>
Heh
14:56
<annevk>
this latest email to webkit-dev...
14:56
<annevk>
MikeSmith is gonna love it
14:56
<gsnedders>
jgraham: The parent of the first commit on the branch? So the merge-base?
14:56
<jgraham>
gsnedders: It's the SHA1 of the parent of the first commit on the branch
14:57
<jgraham>
gsnedders: Yes.
14:57
<gsnedders>
jgraham: Okay, done
14:57
<jgraham>
gsnedders: Basically critic constructs a diff of the post-rebase branch compared to the pre-rebase branch
14:57
<jgraham>
So you need to tell it where the post-rebase branch starts
14:58
<gsnedders>
jgraham: Note that this still fails a whole load of tests, but mostly because the tests are kinda broken
14:59
<tobie>
annevk: np. thanks for merging it (SW patch).
15:03
<gsnedders>
jgraham: Why is the review not tracking any more?
15:05
<Domenic>
I didn't realize that fetch() accepted a request. in that case there should almost certainly be a Request.prototype.send() for when you already have a Request object
15:05
<jgraham>
gsnedders: You need to reenable that
15:05
<Domenic>
fetch(), wherever it ends up, seems mostly for the URL-accepting case to me.
15:05
<tobie>
annevk, seems you didn't push it to gh-pages though.
15:07
<tobie>
annevk: which reminds of https://github.com/slightlyoff/ServiceWorker/issues/266
15:08
<annevk>
Domenic: you want Request for all the additional parameters
15:09
<gsnedders>
jgraham: I pressed the button. Nothing happeend.
15:09
<Domenic>
annevk: you could add those as an options object to fetch, hmm. fetch(url, { timeout: 5000 }) vs. (new Request({ url: url, timeout: 5000 }).send()
15:09
<Ms2ger>
gsnedders, force-refresh
15:10
<Domenic>
woah why is there a synchronous flag O_O
15:10
<jgraham>
gsnedders: Force reload
15:11
<jgraham>
gsnedders: Then blame jl
15:22
<gsnedders>
I blame jl.
15:53
<annevk>
Domenic: I think I filed a bug on that, if that's exposed it should only be readonly
15:55
<Domenic>
annevk: when is it useful?
15:55
<annevk>
Domenic: I guess a service worker might want to know about synchronous requests so it can log errors somewhere for the frontend team
15:55
<Domenic>
annevk: ah right i forgot to context switch from fetch to SW's onfetch
16:22
<annevk>
Domenic: I think the idea for fetch() was String or Request
16:22
<Domenic>
annevk: yeah, I guess, just thinking of what people are used to from jQuery etc.
16:22
<annevk>
Domenic: or a dictionary
16:22
<Domenic>
it's a small step from dictionary to string + dictionary ;)
16:22
<annevk>
fetch({url:...}) should work I think
16:23
<Domenic>
the nice thing about fetch(url, {...}) is that it's an easy change from code that does fetch(url)
16:23
<jgraham>
Putting a non-optional parameter into a dict seems ugly
16:24
<Domenic>
that too
16:25
<annevk>
I guess that's fair, ideally we align the Request constructor with that pattern
16:25
<annevk>
I should probably take ownership at some point, not moving very quickly at the moment
16:27
<tobie>
node.js request accepts both req(url, options) and req({url: url, options... })
16:27
<tobie>
fwiw
16:28
<Ms2ger>
That seems somewhat unhelpful
16:30
<tobie>
Ms2ger: people tend to have strongly diverging opinions when it comes to API design.
16:32
<Domenic>
jQuery also accepts both
16:32
<Domenic>
i kind of dislike the both approach also, but not strongly
16:45
<JakeA>
Domenic: annevk: fetch(url) is not as common as maybe originally thought. I'm happy to ditch fetch() for request.send()
16:46
<JakeA>
As long as new Request(url) works
16:46
<Domenic>
JakeA: hmm why is that? $.get(url) is very common...
16:46
<JakeA>
Domenic: actually, I was thinking of ServiceWorker, in a page fetch(url) would be common
16:47
<JakeA>
But new Request(url) seems ok to me. Happy to be wrong though. Feels like a smaller footprint for the window object
16:50
<annevk>
JakeA: we can do fetch(url, rest); new Request(url, rest); fetch(Request); and maybe new Request(Request) (for downgrading a UA-generated object)
16:52
<jsbell>
annevk's suggestion is what I had in my head (apart from Request(Request) but that makes perfect sense)
16:58
<JakeA>
So function fetch(args...) { return new Request(arts)}
16:58
<JakeA>
Ffs, coding on phone
16:58
<JakeA>
So function fetch(args...) { return new Request(args...).send(); }
16:59
<Ms2ger>
...args?
16:59
<JakeA>
Yeah, that, probably (can never remember which side the ... goes on)
17:02
<jsbell>
If it was exactly that, it auto-downgrades a UA-generated object.
17:03
<Domenic>
I like that
17:05
<jsbell>
BTW, is there a doc/thread yet, or just noodling here? (I'm catching up on email after being away for a bit, so may discover it shortly)
17:12
<annevk>
jsbell: noodling here
17:12
<jsbell>
thx
17:12
<JakeA>
What do we mean by downgrades?
17:13
<JakeA>
That happens on .send() right? It becomes "connect" in CSP terms
17:32
<annevk>
JakeA: yeah, I was thinking we could offer explicit syntax for it as well so people can reason about it without doing a fetch
17:32
<annevk>
JakeA: I'm not sure I like the .send() design
17:37
<annevk>
On the other hand, if we name it .send() there's less of a confusion between fetching and fetch()
17:46
<Hixie>
zcorpan: i figured i would filter the incoming bugs for you rather than just automatically send img bugs to you
17:46
<Hixie>
TabAtkins: autorotate="" is in zcorpan's area now, unfortunately, but maybe i can send him a patch or something
17:47
<Hixie>
as far as the fading thing goes, yeah, i don't like it either
17:47
<Hixie>
i'm trying to figure out how to address the feedback that some people have asking for the spec to be less busy, while still addressing the needs of people like me who use the styles to understand what's going on.
17:48
<annevk>
and here I thought my browser had a bug
17:50
<jgraham>
Uh yeah. That seems like a bad idea
17:50
<jgraham>
Why not just drop the underlines?
17:52
<Hixie>
there are two groups of people i'm trying to cater for. group A wants no colour and no underlines. Group B wants colour and underlines.
17:53
<Hixie>
(group A basically wants no hyperlinks visible, the point as far as i can tell is to feel like you're reading a book rather than feel like you're reading a spec with deep links everywhere.)
17:53
<Hixie>
(group B might be just me.)
17:54
<jgraham>
Since I apparently don't fall into either group, I doubt your characterisation is accurate
17:55
<Hixie>
well it's quite likely there are other groups that i should also be trying to cater for, but those were the groups i was trying to cater for.
17:56
<boogyman>
Hixie: I don't think it would be terribly difficult to use alternate stylesheets, right? The question there though is what interface to use to switch between them.
17:57
<jgraham>
I think that hypertext is a great thing and I think it is especially great in the spec, since you basically can't understand it without following links. So I think the "Group A" people have either been mischaracterised or don't really want what they say they want. OTOH I think that underlines are a particularly poor typographical technique since they compete with the letters themselves
17:57
<Hixie>
boogyman: most users don't touch that kind of UI
17:57
<Hixie>
boogyman: so it doesn't solve the problem
17:57
<IZh>
Hixie: Hi. What's the purpose of lots of empty <span></span> in the spec?
18:00
<jgraham>
http://www2003.org/cdrom/papers/refereed/p391/p391-obendorf.html seems relevant
18:00
<Hixie>
jgraham: the feedback specifically is things like "i wish it didn't look like a christmas tree", "too bright and contrasty", "don't like the colour formatting", "excessive hyperlinks make it too busy", "it's a little cluttered and busy", "it's not the prettiest thing in the world", "dislike layout & font", "looks cheap", "ugly colors", "when everything is highlighted italic red green or blue it is hard to distinguish content"
18:00
<jgraham>
"Underlined links seem to substantially decrease the reading performance on Web pages and may add to the reasons why users donít like to read on the Web"
18:02
<jgraham>
Hixie: Those are all arguments in favour of more subtle design, not in favour of removing the most important elements needed to navigate the spec.
18:02
<Hixie>
(not clear that whoever formatted that page is allowed to comment on formatting, but... *reads*)
18:02
<Hixie>
jgraham: concrete suggestions welcome
18:05
<jgraham>
Hixie: Well so far I made one
18:06
<Hixie>
"To reduce the influence that different degrees of interest in the test items would have, we selected a very homogeneous user group. The target group consisted of regular and experienced internet users, as we wanted to assess the willingness of these users to adopt changes in the Web interface."
18:06
<Hixie>
+1 for making a plausible argument for why they only selected people within shouting distance of their office :-)
18:07
<IZh>
Hixie: Sorry. It seems there are no in latest version.
18:07
<Hixie>
IZh: hey, sorry, missed your question. dunno what could cause that, but probably a markup error on my side.
18:10
<jgraham>
Hixie: I also think that the green and the blue and the yellow don't really go together. The green italic text is quite hard to read, and I wonder if CSS-style boxes with a background colour and black text would work better for notes (more like examples)
18:11
<jgraham>
I am terrible at design though so I am really the wrong person to fix things
18:12
<IZh>
Hixie: please look at 4.12.4.2.7 at definition list after "Run the appropriate step..." It consists of only <dd> and no <dt>. Is it correct?
18:13
<Hixie>
jgraham: that paper's interesting, but the conclusion for the spec isn't to get rid of underlines and even less to make hyperlinks only visible on demand, imho. Assuming we treat reading the spec as more like their "link" tasks, it would suggest hyperlinks should be always visible (fewer errors in that case), and looking at the feedback of their overlay vs underline thing, it seems like a toss-up regarding which people like best.
18:13
<Hixie>
jgraham: yeah, same here
18:13
<Hixie>
jgraham: what's yellow?
18:14
<Ms2ger>
The sun?
18:14
<Hixie>
in the spec, doofus
18:14
<Hixie>
IZh: what's the heading of that section?
18:14
<Hixie>
IZh: (i don't have section numbers in the source)
18:15
<IZh>
Hixie: Path2d objects
18:16
<Hixie>
that's weird, wonder why the validator didn't catch that
18:17
<Hixie>
IZh: fixed, thanks
18:17
<IZh>
Hixie: I have bought one commercial validator. And of course couldn't test it against the spec. ;-)
18:18
<annevk>
There's commercial validators now?
18:18
<jgraham>
Hixie: Links on hover are yellow. And link targets
18:18
<IZh>
Hixie: I mean couldn't not test ;-)
18:19
<IZh>
annevk: Yes. There are some good.
18:19
<Hixie>
jgraham: oh the hover effect, ok
18:20
<annevk>
Hixie: I like the new Example / Note / Warning / etc. thing
18:21
<jgraham>
Hixie: I think the conclusions of the study are a) too suble link styling makes people not use links and b) too invasive link styling makes text hard to read. I think that dropping the underlines will help with linear reading
18:21
<IZh>
Hixie: by the way, what is the source for the spec? Xml?
18:21
<jgraham>
I doubt it will make the links invisible, so I don't think we'll hit the error case
18:22
<Hixie>
IZh: some weird variant of HTML with preprocessor directives
18:23
<Hixie>
jgraham: they specifically say in the study (without data sadly) that they think that links that are only distinguished by colour are too subtle (they think it would be the same as tplain text), so that would be case (a)
18:24
<IZh>
Hixie: validator also complains about very long lines. There is, for example, the table of named entities, that is only single line.
18:24
<jgraham>
Hixie: They say without data is about as valuable as I say the opposite without data :)
18:24
<jgraham>
Except that they are presumably trying to justify not taking the time to test that case
18:25
<Hixie>
IZh: that
18:25
<Hixie>
IZh: that's an error in the validator :-)
18:25
<Hixie>
jgraham: exactly :-)
18:26
<IZh>
Hixie: it only complains about syntax highlighting. Also is is not very huumsn
18:26
<IZh>
Hixie: human readable
18:26
jgraham
discovers that http://usability.gov doesn't underline links
18:27
<jgraham>
Actually I think the fact that so many sites now don't underline links sort of puts the anecdata on my side of the argument
18:27
<IZh>
Hixie: and some browsers performed bad when you try to view source of the document with such long lines.
18:29
<Hixie>
i find sites that don't underline links very confusing, personally
18:31
<annevk>
I think I agree with jgraham
18:32
<annevk>
color is quite a good indicator and the underline makes the text harder to parse
18:37
<Hixie>
so without the underline, how do you distinguish a <code> snippet that's a link from one that isn't?
18:38
<IZh>
Hixie: one long line contains 314187 characters. :-)
18:38
<Hixie>
IZh: yeah, it's machine-generted :-)
18:42
<IZh>
Hixie: the links to [CSSFONTLOAD] and [PAGEVIS] at the end of the spec are 404.
18:42
<annevk>
Hixie: maybe that should have the color of :link / :visited and simply use monospace
18:42
<Hixie>
annevk, sicking: ok, reload a whatwg spec, tell me how bad it is...
18:42
<annevk>
Hixie: :hover / :focus could add the underline
18:43
<Hixie>
IZh: sigh, w3c
18:43
<annevk>
Hixie: in http://url.spec.whatwg.org/#dom-url TypeError is grey
18:43
<Hixie>
anyone know the currently URLs of http://dev.w3.org/csswg/css-font-load-events/ and http://www.w3c-test.org/webperf/specs/PageVisibility/ ?
18:44
<Hixie>
annevk: yeah, that's a non-hyperlink, non-definition <Ccode> block.
18:44
<Hixie>
<code> even
18:44
<annevk>
so only code that's hyperlink or <dfn> is orange?
18:44
<IZh>
Hixie: Is this it? http://dev.w3.org/csswg/css-font-loading/
18:45
<annevk>
TabAtkins: ^^ might want to set up redirects
18:45
<IZh>
http://dev.w3.org/csswg/css-font-loading/
18:45
<IZh>
http://dev.w3.org/csswg/css-fonts-3/
18:46
<annevk>
Hixie: there's http://www.w3.org/TR/page-visibility/ can't find editor's draft :/
18:46
<Hixie>
IZh: none of those seem to define FontLoader... i wonder if FontLoader is dead or something
18:46
<Hixie>
annevk: currently, yeah
18:46
<Hixie>
annevk: i've no idea if that's reasonable
18:46
<annevk>
Hixie: FontLoader is dead I think
18:47
<Hixie>
huh
18:47
<Hixie>
guess i'd better file a bug to handle that...
18:47
<annevk>
Hixie: no editor's draft would mean no maintenance
18:47
<annevk>
Hixie: http://dev.w3.org/csswg/css-font-loading/ has an alternative AIP
18:47
<IZh>
Hixie: http://www.w3.org/TR/2012/WD-css3-fonts-20121211/ The old one defines it.
18:47
<annevk>
API*
18:47
<Hixie>
file a bug
18:47
<Hixie>
er
18:47
<Hixie>
fileD a bug
18:47
<Hixie>
https://www.w3.org/Bugs/Public/show_bug.cgi?id=25812
18:48
<Hixie>
IZh: (thanks for finding these!)
19:12
<IZh>
Hixie: Is it good or bad when quoted string attributes spans across several lines?
19:16
<IZh>
Like: <a href=#syntax title="the
19:16
<IZh>
HTML syntax">HTML syntax</a>
19:27
<IZh>
There are some of these in the spec.
19:31
<caitp>
it works in current browsers, so presumably the parsing spec allows for it
19:31
<caitp>
unless it's just a happy accident
19:36
<IZh>
Hixie: In the end of the section "The WorkerGlobalScope common interface" there is empty link <a href=#dom-workerglobalscope-location></a> before the word "attribute".
19:37
<IZh>
Hixie: The same thing in the section "Standard metadata names": using the <code title=attr-lang><a href=#attr-lang></a></code> attribute...
19:39
<Hixie>
izh: looking...
19:40
<Hixie>
TabAtkins: what's the state of the art with respect to positioning something relative to an ancestor element (e.g. one with position:relative) for 'top', and relative to another (e.g. the root element or ICB) for 'left'?
19:41
<Hixie>
IZh: thanks, fixed
19:41
<SamB>
Hixie: don't you need nasty hacks for that?
19:42
<Hixie>
SamB: that's what i'm asking
19:42
<Hixie>
so far for elaborate cases i'm not finding any real solutions
19:42
<Hixie>
it's even worse because i have some elements with position:relative for unrelated reasons, too
19:43
<IZh>
What's the purpose of &blank;? How it differs from ordinary space?
19:44
<Hixie>
it's visible, for one
19:44
<Hixie>
it's used to show that there is a space, in code examples where spaces matter
19:44
<Hixie>
(this glyph used to be a lot more commonly used in computer manuals from a few decades ago)
19:45
<IZh>
Ahh... Sonething like '_' ?
19:45
<Hixie>
yeah
19:45
<Hixie>
jgraham: i've changed the :target style to not be yellow too.
19:52
<IZh>
I'm curious, will ever document title support markup? ;-) I want green window caption. ;-)
20:03
<Hixie>
IZh: given that document titles are getting less and less used in the UI, i doubt there's much demand to make them harder to implement :-)
20:07
<jgraham>
Hixie: I think that's a win
20:09
SamB
is greatly saddened by the disuse of document titles :-(
20:10
<SamB>
however, it's not bloody likely that they'll be getting any such fancy features
20:11
<SamB>
the places where they probably still are used -- normal WMs on *nix come to mind -- can't handle anything fancy anyway
20:12
SamB
sorta feels like it might be nice if stray tags were stripped rather than shown verbatim though
20:17
<IZh>
One more question. Is it possible to change facicon ob the fly?
20:20
<SamB>
IZh: well, what happens if you change the relevant link element(s)?
20:21
<IZh>
SamB: perhaps it will work.
20:31
<IZh>
is available. And in the case of no connection it will display an error instead of old cached version. I spend lots of time in a waiting for cellular network to appear. Probably some pragmas could force the browsers to display old cached page when they can't fetch a new version instead of showing dumb error.
21:33
<JonathanNeal>
How are element queries coming along?
22:28
<TabAtkins>
Hixie: What is FontLoader and what were you using it for? If you just want to load a font via JS, FontFace is constructable.
22:28
<TabAtkins>
Hixie: There is no way to position the top and left of an element relative to different things at the moment.
22:39
<Hixie>
TabAtkins: i don't recall. something in HTML. i filed a bug on myself to figure it out.
22:39
<Hixie>
TabAtkins: re positioning: k. anything on the radar for that?
22:47
<JonathanNeal>
To whomever it concerns, I have filed my latest element query pleas @ http://discourse.specifiction.org/t/element-queries/26/last
22:54
<TabAtkins>
Hixie: Nope, nothing right now. I have plans, but no timeline for achieving them.
22:54
<Hixie>
k
22:56
<Domenic>
http://gridstylesheets.org/ is a related prolyfill/ideation exercise
22:57
<Domenic>
i want to try it on a small project and see how i feel afterward
22:57
<Domenic>
i could easily see it either being "wow this is amazing we must have this" or "meh that really doesn't work in practice, nice try"
23:09
<JonathanNeal>
Domenic: were you the one who replied to me?
23:10
<JonathanNeal>
If it was, I appreciate you citing TabAtkins’ blog as the canonical reply. But, at the same time, arrrrr! I had a bunch of links in my post, but discourse doesn’t let me add more than one or two.
23:18
<zewt>
are clipboard events those sort of broken ancient events that have side-effects when fired from script (like click)?
23:19
<zewt>
it doesn't seem like they are in testing, but hallvord implied that they are on the list (and he's apparently the editor of the clipboard spec)
23:24
<JonathanNeal>
Domenic: re: gss, it’s amazing that we can think up entirely different ways to write stylesheets, and its best marketing is still “now you can vertically center something”
23:31
<gsnedders>
Okay then, to continue my questioning from yesterday: which Chromebook do I want? Given they basically all have 1366x768 displays, I guess there's no point in going for one of the larger ones? AFAICT, the HP Chromebook 11 seems to have the best screen of them, but it does have a now ancient SoC in it…
23:38
<JonathanNeal>
gsnedders: that sounds all right. I’m very productive on 1440x900.
23:41
gsnedders
vaguely wonders about selling his tablet
23:41
<gsnedders>
I just don't use it that much…
23:50
<JonathanNeal>
gsnedders: i used mine for one project, and my 1 year old found a better use for it, namely the tinkerbell film series.
23:52
<gsnedders>
All the Chromebooks seem to be somewhat flawed. The HP Chromebook 11 has a wonderful screen, but opinions are mixed on the keyboard, and apparently performance isn't brilliant with a few tabs open… On the other hand, the Dell Chromebook 11 has a less good screen but better everything else…