03:59
<MikeSmith>
mkwst: FYI https://stackoverflow.com/questions/38176441/block-all-mixed-content Wondering what the current browser support is for CSP block-all-mixed-content. I thought Firefox and Chrome both already supported it at least, but comment there is “Unfortunately block-all-mixed-content doesn't work in Firefox”
04:00
<MikeSmith>
mkwst: and please post an answer or comment to that SO question if you have other info as far as “Is there a cross-browser way”
07:20
<mkwst>
MikeSmith: I'm pretty sure Mozilla folks implemented it. /cc dveditz_.
07:21
<MikeSmith>
k
07:23
<dveditz_>
https://bugzilla.mozilla.org/show_bug.cgi?id=1122236 -- it's in Beta now (Firefox 48)
07:24
<dveditz_>
MikeSmith: ^^^^
07:24
MikeSmith
looks
07:25
<MikeSmith>
w00t
07:25
<MikeSmith>
dveditz_: thanks!
07:25
<dveditz_>
(and now off to bed for me)
07:59
<MikeSmith>
zcorpan: FYI https://lists.w3.org/Archives/Public/www-validator/2016Jul/0001.html somebody saw the new lnag warning and tried doing <!DOCTYPE html lang="en"> to fix it
08:00
<MikeSmith>
zcorpan: A tradeoff with the way I ended up implementing this to get around the other earlier problem is that now I unfortunately cannot report the line+column number of the html element in the warning message
08:02
<zcorpan>
MikeSmith: :-| i guess "the html element start tag" is too much jargon? Maybe add an example with doctype + html start tag in the message?
08:03
<MikeSmith>
yeah
08:03
<MikeSmith>
I suppose so
08:03
<MikeSmith>
will change it to that
08:03
<zcorpan>
do you have the html token available to build the message?
08:04
<MikeSmith>
yes
08:04
<MikeSmith>
with the attributes
08:05
<MikeSmith>
though not the lexical bits, like whether the attribute values are quoted or not
08:05
<MikeSmith>
or whitespace/newlines inside the start tag
08:06
<zcorpan>
sure, but would be cool to have the right attributes in the example :-)
08:07
<zcorpan>
another thing i noticed when looking at httparchive examples was that pages that use html5boilerplate with IE conditional comments around <html> tag sometimes neglect to change the langs for IE
08:41
<MikeSmith>
zcorpan: oh yeah the other day you pointed me at an example of that I think
08:42
<MikeSmith>
I hope I don’t need to end up implementing a comment parser to catch and report those...
08:45
<annevk>
MikeSmith: planning on updating https://github.com/whatwg/html/pull/1470?
08:48
<MikeSmith>
annevk: yes, will do right now. Thanks for finishing the review and sorry for not getting back it yetーbeen working on this lang-detecting feature in the HTML checker.
08:48
<g4>
does anyone here know if there is a way to detect if a IME is open or not? e.g. on a keydown event?
08:48
<MikeSmith>
g4: kochi might know
08:51
<annevk>
MikeSmith: sounds like that would be more fun; no worries
09:06
<zcorpan>
MikeSmith: http://stackoverflow.com/questions/38178184/this-document-appears-to-be-written-in-french-w3-validator-warning
09:08
<zcorpan>
MikeSmith: https://forums.adobe.com/thread/2175784
09:10
<zcorpan>
i suppose with 99.something confidence it will get it wrong in 0.something of cases
09:11
<MikeSmith>
annevk: done I think, review fixes for https://github.com/whatwg/html/pull/1470
09:11
MikeSmith
looks at zcorpan links
09:12
<zcorpan>
on the good news side https://github.com/sumBTC/TheBitcoinEcosystem/commit/88e3adc5a6eedeeef467ced8795a523c6c830237 🙂
09:16
<MikeSmith>
zcorpan: about the false-positive possibility, yeah I was aware that could happen, especially with documents that aren’t sentences and paragraphs of text but instead just menus or links
09:17
<MikeSmith>
the library has an API for examining the probability data, so perhaps I will need to implement something a bit smarter using that
09:20
<zcorpan>
MikeSmith: if i specify <html lang=""> i get two warnings. empty string is an allowed value and means language is unknown, but... i suppose nobody deliberately uses it on purpose
09:21
<MikeSmith>
zcorpan: still I should try to fix that
09:21
<zcorpan>
MikeSmith: yeah that seems like a good idea; if I validate a doc with lots of numbers and one actual word, it triggers the warning but probability will be low
09:22
<MikeSmith>
right
09:22
<MikeSmith>
or even just with lots of spacs and newlines
09:22
<MikeSmith>
SAX API does not give a good way to discard whitespace
09:23
<MikeSmith>
I just collect characters regardless
09:24
<zcorpan>
hmm... you just collect the characters and don't particularly care about tags, right? so <p>foo<p>bar will be one word "foobar"?
09:32
<kochi>
g4: within what is available today, you need to handle 'compositionstart' and 'compositionend', and IME is active between the two events.
09:33
<kochi>
g4: depends on what you define as "IME is open", YMMV.
09:33
<MikeSmith>
zcorpan: yeah it would be
09:34
<kochi>
g4: for instance, if you turn on IME on windows, your IME is kind of open, but composition* events happen only after you start actual typing.
09:34
<MikeSmith>
annevk: I think I missed one “creating a potential-CORS request” that I need to add the destination param to
09:34
<annevk>
MikeSmith: ok
09:35
<annevk>
MikeSmith: do you agree that "_destination_ set to the empty string" should become "the empty string"
09:35
<MikeSmith>
annevk: yup, that is more clear, more simple
09:35
<MikeSmith>
already pushed that change just now
09:36
<zcorpan>
MikeSmith: i don't know how much it affects the results, but knee-jerk fix if it messes things up (e.g. for menus with no whitespace between tags because minimizers), is to append a space for a set of block-ish elements like p, li, h1...
09:36
<zcorpan>
for the end tag
09:37
<annevk>
ah I see, looks pretty good now MikeSmith, will wait for that final fix then
09:37
<MikeSmith>
annevk: OK, about that, can you please take a second to look at line 86229 on that branch? the algorithm <h5 id="fetching-scripts">Fetching scripts</h5>
09:39
<MikeSmith>
annevk: it seems like it needs to specify, <var>url</var>, *the empty string*, and <var>CORS setting</var>, right?
09:39
<annevk>
MikeSmith: no
09:39
<MikeSmith>
oh
09:39
<MikeSmith>
good thing I asked
09:39
<annevk>
MikeSmith: it needs to be "script" from a following step
09:39
<MikeSmith>
oh
09:39
MikeSmith
re-reads
09:39
<annevk>
MikeSmith: and you then need to remove that destination setting in the following step
09:40
<MikeSmith>
annevk: oh, maybe will create a separate PR for that then?
09:40
<MikeSmith>
because that is a different change than what we are doing in this PR
09:41
<MikeSmith>
and can just land this one first since it is otherwise OK
09:41
<annevk>
MikeSmith: I think it's better to do it as part of this PR since you change the signature of "create a potential CORS-request"
09:41
<MikeSmith>
ah OK
09:41
<MikeSmith>
fair enough
09:41
<annevk>
MikeSmith: since if you land this patch, the invocation there results in an error theoretically
09:41
<MikeSmith>
right
09:41
<MikeSmith>
yup
09:42
<MikeSmith>
ok then I will need to do that after a bit, since I gotta drop off soon to catch the bus+train back home
09:43
<annevk>
okido
09:48
<MikeSmith>
annevk: thanks for the careful review
09:57
<g4>
kochi: very useful! Did not know about those events. Thanks!
12:17
<annevk>
I missed an "Opera did it first" in https://github.com/whatwg/html/issues/1482
15:01
<zcorpan>
quiz time https://twitter.com/zcorpan/status/749976380247605248
15:04
<Ms2ger>
Dammit
17:18
<Domenic>
what why
17:18
<Domenic>
one day i will get a htmlquiz
17:22
<MikeSmith>
annevk: OK just pushed that one remaining “creating a potential-CORS request” change
17:23
<MikeSmith>
(the the PR branch)
17:46
<annevk>
MikeSmith: \o/
17:47
<MikeSmith>
w00t
17:47
<MikeSmith>
thanks
20:48
<zcorpan>
MikeSmith: https://github.com/validator/validator/issues/315