00:08
<marcosc>
Hixie: I added the basic abuse case - http://w3c-webmob.github.io/wake-lock-use-cases/#potential-for-abuse
00:08
<marcosc>
I'm asking on Twitter for more
00:08
marcosc
heading home now... but if anyone has any others please let me know
00:10
<caitp>
i guess that would be a fun bug for some college kids to work on at least, so it's got that going for it
12:00
<JakeA>
Anyone awake enough to set me up an account on the whatwg wiki?
12:03
<jgraham>
JakeA: Sure I can do it
12:04
<JakeA>
jgraham: Cheers!
12:05
<JakeA>
jgraham: jakearchibald jaffathecake⊙gc
12:31
<MikeSmith>
the "polyglot" document is very consistent about being really sloppy
12:32
<Ms2ger>
Ha
12:33
<MikeSmith>
if you look at the Status section at http://www.w3.org/TR/2014/WD-html-polyglot-20140204/ and find the link with the hypertext "facilitate migration to and from XHTML" and you follow it.. it goes to http://www.w3.org/html/wg/drafts/html/master/document-metadata.html#charset "Specifying the document's character encoding"
12:33
<MikeSmith>
...which says nothing about how to "facilitate migration to and from XHTML"
12:34
<MikeSmith>
and there's a pattern of instances of stuff like that throughout the document
12:37
<Ms2ger>
You're not supposed to click the links
12:42
<MikeSmith>
no, you're supposed to put yourself in the mind of the authors before you click the links
16:04
<Domenic>
Has nobody built something that auto-generates JS function headers from WebIDL?
16:08
<MikeSmith>
Domenic: as opposed to C++ function headers? I would that somebody from a browser project had moved some feature completely to JS might must have
16:08
<MikeSmith>
I mean I think there have been some cases of things being moved to JS
16:09
<Domenic>
MikeSmith: well, I am starting that browser project to move features completely to JS, so... not here at least :P
16:09
<caitp>
do optional parameters in webidl affect the arity of a function / are they supposed to?
16:09
<MikeSmith>
not just abarth's marquee implementation, but other things earlier
16:10
<MikeSmith>
Domenic: maybe something over there in gecko (me waves hands around a bit)
16:20
<jgraham>
We can do you C++ or Rust, but not js
16:28
<Ms2ger>
Domenic, we have JS-implemented stuff, but I don't think anything to generate stub implementations
16:29
<Domenic>
Ms2ger: thanks... I'll probably start something on GitHub then ^_^
16:31
<Domenic>
Oh wait, I guessed a URL and found marcosc's https://www.npmjs.org/package/webidl
16:36
<MikeSmith>
さすがmarcosc
16:44
<TabAtkins>
caitp: Yes, they do affect the arity.
16:44
<TabAtkins>
caitp: ...pretty sure.
16:45
<caitp>
weird, that's pretty different behaviour from the standard library
16:45
<TabAtkins>
?
16:46
<caitp>
from JS builtins
16:46
<TabAtkins>
Is it?
16:47
<TabAtkins>
Really, it depends on whether they translate to JS as just a normal arg (and only affect the arity/type checking), or if they translate as an arg defaulted to undefined.
16:47
<caitp>
well, it probably varies a lot, but it's frequently "this function has 3 parameters, but a length of 1"
16:47
<cryptic>
should you set content-type to "application/csv" for csv files?
16:47
<TabAtkins>
caitp: Oh, you might have misunderstood me.
16:47
<TabAtkins>
I meant that optional args should *not be counted* in the arity.
16:47
<cryptic>
my browser doesn't recognize that mime type, which seems dumb
16:47
<caitp>
ah
16:48
<TabAtkins>
cryptic: Try text/csv?
16:48
<cryptic>
TabAtkins: doesn't text imply ascii?
16:48
<TabAtkins>
No.
16:48
<cryptic>
oh, perfect
16:49
<TabAtkins>
text/html certainly doesn't, for example.
16:49
<TabAtkins>
It just implies that it's a textual format, as opposed to one of the binary formats.
16:49
<cryptic>
but html is ascii
16:49
<TabAtkins>
No, HTML is definitely not ASCII.
16:49
<TabAtkins>
All of the language-defined stuff is ASCII, yes, but HTML pages are not.
16:49
<TabAtkins>
(Rather, is within the ASCII range, which is far different from *being* ASCII.)
16:50
<cryptic>
right, that's what I mean: "text/html" refers to the language of the page content, which is html (ascii)
16:50
<cryptic>
that the browser ultimately does with that information is beyond the jurisdiction of http
16:50
<TabAtkins>
No. The page content is some HTML tags and a bunch of textual content. The latter is not required to be ASCII, and often is not.
16:51
<TabAtkins>
HTML doesn't distinguish between tags and text as being "HTML content".
16:51
<cryptic>
it's all html content
16:51
<TabAtkins>
And HTML tags use characters in the ASCII range of unicode, which may or may not be ASCII-compatible in your chosen encoding.
16:51
<TabAtkins>
cryptic: I know it's all HTML content. That's what I"m saying.
16:54
<cryptic>
at the end of the day, it doesn't matter between text/csv and application/csv; I'm just not sure which one comports with the specifications (that some browsers correctly display unescaped unicode characters is not behavior the specs would lead one to expect)
16:55
<cryptic>
but I guess if text meant ascii, it'd be ascii/* and not text/*
16:56
Ms2ger
notes the topic
16:58
<gsnedders>
Defining "text" is hard :)
17:06
<TabAtkins>
cryptic: If you've been specifying an ASCII charset and escaping all your non-ASCII character in HTML all this time, just in case someone was following some obsolete directive from some ancient version of the text/* definition, then I"m very sorry. ^_^
17:07
<gsnedders>
TabAtkins: hey, Hixie publishes the spec as US-ASCII!
17:07
<TabAtkins>
...why?
17:07
<Ms2ger>
So that he can push it to the IETF, duh
17:07
<gsnedders>
he got fed up of having issues with dev.w3.org and Content-Type headers specificing charset
17:08
<gsnedders>
IIRC
17:08
<gsnedders>
and just changed it to use US-ASCII everywhere
17:08
<TabAtkins>
Oh, huh. I have no idea what those issues might be, since we all use utf-8 in CSS.
17:08
<gsnedders>
source is UTF-8, but the published specs are ASCII
17:08
<gsnedders>
idk, this was years ago this was chnged
17:08
<gsnedders>
*changed
17:08
<TabAtkins>
That's silly.
17:09
<Ms2ger>
W3C? It happens
17:09
<Hixie>
gsnedders: actually source is ascii but i now publish as utf-8
17:09
<Hixie>
then again, i don't publish on the w3c site anymore
17:11
<gsnedders>
Hixie: well you parse source as utf-8 from memory! :P
17:11
<gsnedders>
or rather used to
17:11
<Hixie>
i do parse as utf-8
17:11
<Hixie>
i mean, utf-8 is a superset of ascii, so...
17:11
<Hixie>
anyway, i don't use anolis anymore
17:12
<Hixie>
and once i've written my splitter, i won't use html5lib at all anymore :-)
17:12
gsnedders
remembers he was meant to see how fast Hixie's parser was
17:13
<Hixie>
oh, yeah
17:13
<Hixie>
how should we do that?
17:14
<Hixie>
can i get you a binary or something?
17:14
<Hixie>
what should it do?
17:14
<Hixie>
or do you want to compile from source?
17:14
<gsnedders>
well I'm on holiday now with nothing more than a kinda weak laptop, so let's not worry for now
17:14
<Hixie>
k
17:15
<gsnedders>
also what seems like a ridiculously slow internet connection :)
17:37
<Hixie>
ok i'm not entirely done yet but this patch is getting out of hand so i'm going to commit it soon
17:37
<Hixie>
anyone see any blockers? http://www.whatwg.org/specs/web-apps/current-work/
17:38
<Hixie>
(if you see them, mention them here. i'm going to be afk for about an hour, then i'll commit it)
18:13
<Hixie>
no comments, huh
18:13
<Hixie>
ok
18:27
<Ms2ger>
Hixie, my browser still doesn't like the single-page spec :)
18:27
<Hixie>
yeah
18:27
<Hixie>
that's unlikely to change
18:28
<Hixie>
in fact the pipeline change made it quite a bit fatter
18:28
<Hixie>
since i have way more IDs now
18:30
<jgraham>
The spec's not done until Mozilla* won't run? [*on Ms2ger's laptop]
18:31
<Ms2ger>
I wonder what it does in Servo
18:37
<gsnedders>
Hixie: can I bat my eyelids and ask for ids on all parse errors again? ;P
18:58
GPHemsley
wonders aloud (again?) about whether there will ever be an Acid4 test
18:59
<Ms2ger>
No
19:07
<caitp>
so end-tags can have attributes?
19:07
<caitp>
huh
19:17
<Hixie>
gsnedders: unique unchanging ones?
19:18
<Hixie>
gsnedders: what's the use case again?
19:21
<gsnedders>
Hixie: yeah; for the sake of html5lib-tests being able to assert the correct parse-error is being hit
19:21
<gsnedders>
instead of just "a parse error was hit here"
19:22
<Hixie>
is that a good thing? i mean, we don't want to make implementations have to track exactly what parse error was hit
19:22
<Hixie>
it prevents quite a few interesting optimisations
19:22
<Hixie>
e.g. several times i coallesced parse errors
19:22
<gsnedders>
AFAICT you either want to track them exactly or ignore them entirely
19:22
<Hixie>
nah
19:23
<Hixie>
the point of parse errors is to tell the author that there's a problem
19:23
<Hixie>
the author doesn't care whether it came from section 12.6.34.2.4 or section 12.6.33.1.6.4
19:23
<Hixie>
the author just wants to know that their < is missing
19:23
<Hixie>
or whatever
19:24
<Domenic>
TabAtkins: fix the missing </pre> in http://dev.w3.org/csswg/css-color/ so I can tweet about how cool it is?
19:24
<gsnedders>
so html5lib-python I think uses strings that aren't actually entirely unique to deal with stuff like this
19:24
<gsnedders>
so maybe we don't want actual ids
19:24
<gsnedders>
maybe an @data-parse-error='missing <' or smth
19:25
<TabAtkins>
Domenic: Where is it?
19:25
<TabAtkins>
Whoops, it's a missing <code> I think.
19:25
<TabAtkins>
</code>, rather.
19:26
<Hixie>
gsnedders: again, i don't think that works. There are places where I just fire "something's wrong" and it could be any number of parse errors.
19:26
<Hixie>
gsnedders: other places the spec has one parse error line, but the message should vary dramatically in an implementation giving useful advice
19:28
<gsnedders>
I want something that makes sense to test that the implementation does useful advice. I don't care how we do this!
19:28
<TabAtkins>
Domenic: fixed
19:28
<Domenic>
TabAtkins: ah right yeah, code makes more sense
19:29
<Domenic>
TabAtkins: still there "custom stringifiers on the stringifiers object..."
19:29
<TabAtkins>
Sorry, forgot to commit.
19:29
<Hixie>
gsnedders: i'm not sure how the spec can help there
19:29
<TabAtkins>
It was fixed in my local copy, I don't see what the problem si.
19:30
<Hixie>
gsnedders: "useful" is something that is too subjective
19:40
gsnedders
concludes playing Super Mario Bros is more fun than working out what "useful" is :)
20:13
<Domenic>
Ms2ger: are these the only tests that exist for select? https://github.com/w3c/web-platform-tests/tree/175953ab7f14aedbcf2a5219ede4387bc86a3b91/html/semantics/forms/the-select-element
20:13
<Ms2ger>
Possibly
20:15
<gsnedders>
does anyone have data for how common with/direct-eval are in JS on the web as a whole?
20:15
<caitp>
can we add a test that asserts one particular way to behave when options are dynamically added to a select control?
20:15
<TabAtkins>
Ask on es-discuss, I'm sure someone's got some data.
20:15
<caitp>
i want browsers to do something consistent there :[
20:16
<gsnedders>
TabAtkins: bah, I'm not subscribed to es-discuss any more!
20:17
<jgraham>
gsnedders: Seems unlikely? Well unless people are instrumenting actual js engines I suppose. Are they?
20:17
<Ms2ger>
Good call
20:20
<gsnedders>
jgraham: you could do it at an AST level, and you guys definitely have some instrumentation in SM
20:20
<gsnedders>
jgraham: so you can do it cheaply
22:11
<smaug____>
dglazkov_: so do you or someone else from blink team have plans to go through all the in-doc vs. not-in-doc cases and file spec bugs when stuff should work also in shadow DOM ?
22:11
<smaug____>
currently implementing shadow dom in interoperable way isn't really a trivial task
22:12
smaug____
will be going through about 800 cases when gecko uses IsInDoc or GetCurrentDoc
22:14
<smaug____>
oh, it wasn't quite that much
22:14
<smaug____>
perhaps only 400
22:17
<mitchell>
Hello. Would you guys please help me find a list or table of 63 established *-English Braille symbols (1 cell, 6-dots) and their correlating ASCII characters?
22:19
<mitchell>
for instance, http://www.duxburysystems.com/braillechart.asp but limited to one cell
22:20
<mitchell>
actually, words 63 words instead of characters would be okay too
22:37
<TabAtkins>
mitchell: You looking for the character information? Here's all of Braille in Unicode: http://symbolcodes.tlt.psu.edu/bylanguage/braillechart.html
22:37
<TabAtkins>
But that doesn't help with the definitino of them. What are you having trouble finding, precisely?
22:46
<mitchell>
TabAtkins: yea, I need an alphabet which gives meaning to those symbols. my goal is write a 2 character identifier on ~2,000 pieces of paper in both English and Braille.
22:47
<mitchell>
s/alphabet/dictionary
22:47
<TabAtkins>
What's lacking in http://www.duxburysystems.com/images/bana_black.pdf ?
22:49
<mitchell>
I thought it was what I needed at first, and maybe I'm misunderstanding it, because it seems to use multiple cells
22:49
<TabAtkins>
For some words, yes.
22:49
<TabAtkins>
But the letters are single-cell.
22:49
<mitchell>
so, for example "child". that is 1 cell, and "children" is 2?
22:51
<mitchell>
I think I'm getting a little more clear with it. I think I'll just need to cross out the multi-cell terms and then it'll be good, hopefully
22:51
<TabAtkins>
The word "child" appears to be, yes.
22:52
<TabAtkins>
Note that "child" apparentlyl has the special rule that it must be surrounded by spaces.
22:52
<mitchell>
whoa, good point
22:54
<jsbell>
nifty, didn't know Braille had an extra layer
22:54
<mitchell>
the pattern that I'm seeing is that I'm basically picking up 2-letter combinations with the remaining 36 (63-27) symbols
22:54
<mitchell>
jsbell: what do you mean by layer?
22:55
<jsbell>
That it's not just 1:1 with English glyphs.
22:55
<TabAtkins>
mitchell: Yeah, it seems like the remaining symbols are for word shortening.
22:56
<jsbell>
Like a standard set of abbreviations that include special use non-letter "glyphs"
22:56
<jsbell>
I wish I was young enough to learn Classic Mayan script. :P
22:56
<caitp->
you are!
22:57
<mitchell>
jsbell: yea, shortened words seems to be common across different Braille systems (as far as my reading today shows)
22:57
<mitchell>
TabAtkins: thanks for helping me understand this chart better
22:58
<TabAtkins>
mitchell: No problem.