09:46
<odinho>
Grrr. Mercurial. Hate, hate.
09:46
<odinho>
Y U NO rebase? :|
09:47
<hsivonen_>
odinho: it has rebase, but last I heard, the advice was not to use it because of a catastrophic bug
09:47
<hsivonen>
I’m rather unhappy about the impact of ambiguous ampersand checking
09:51
<zcorpan>
hsivonen: impact being that less cases are flagged by the validator?
09:55
<odinho>
Finally able to push. Phew. hg always manages to make me grumpy. Have to do something fun now. Lunch I guess :P
09:55
<hsivonen>
zcorpan: the impact being having to touch the most brittle part of the tokenizer code
09:55
<zcorpan>
hsivonen: ah
10:23
<zcorpan>
TabAtkins: re https://bugs.webkit.org/show_bug.cgi?id=39140 doesn't webkit already use a different parser for html colors?
12:37
<darobin>
so for a redirect on a DELETE, Firefox prompts (and does it if you accept the prompt), Safari and Opera automatically fail (they send a GET to the redirect URL), Chrome automatically redirects as a DELETE
12:37
darobin
wonders if anyone is concerned with HTTP interop
12:49
karlcow
is not concerned but interested by HTTP interop ;)
13:08
<zcorpan>
darobin: have a TC?
13:09
<zcorpan>
darobin: is this xhr?
13:09
<darobin>
zcorpan: yup, gisting TC, gimme a sec
13:11
<darobin>
zcorpan: https://gist.github.com/3852447
13:15
<darobin>
zcorpan: note that when I say "fail" above, I mean "fail against what I'd like them to do" — it may well be that some exegesis of the HTTP spec could construe that as the correct behaviour
13:16
<zcorpan>
xhr spec has a note saying "Note: HTTP places requirements on the user agent regarding the preservation of the request method and request entity body during redirects, and also requires end users to be notified of certain kinds of automatic redirections."
13:17
<zcorpan>
http says "The action required MAY be carried out by the user agent without interaction with the user if and only if the method used in the second request is GET or HEAD."
13:18
<zcorpan>
(for 3xx)
13:18
<zcorpan>
for 302, "If the 302 status code is received in response to a request other than GET or HEAD, the user agent MUST NOT automatically redirect the request unless it can be confirmed by the user, since this might change the conditions under which the request was issued."
13:20
<karlcow>
http://tools.ietf.org/html/draft-ietf-httpbis-p2-semantics-21#section-7.4
13:20
<zcorpan>
same with 307
13:24
<zcorpan>
well, after reading the spec, i have no idea what's supposed to happen
13:24
<darobin>
zcorpan: that's pretty much where I ended up after trying to read it :)
13:25
<darobin>
a strict reading might give Firefox as the one right way, but frankly prompting people to accept a redirect for an XHR request is UI madness
13:27
<zcorpan>
http://trac.tools.ietf.org/wg/httpbis/trac/report/9 seems to be the open non-editorial issues
13:29
<zcorpan>
i guess filing an issue "actually define what is supposed to happen" is not likely to do the trick
13:30
<zcorpan>
darobin: browsers could refuse the redirect without prompting
13:30
<darobin>
zcorpan: that's what Opera and Safari do, but it's not really all that useful either
13:31
<zcorpan>
darobin: i thought you said they change to GET
13:31
<zcorpan>
that's not the same as not redirecting
13:31
<darobin>
oh, yes, you actually not do anything — yes, indeed
13:32
<darobin>
in Firefox if you cancel the prompt that's what it does, you get the redirect response directly in XHR
13:32
<zcorpan>
but i don't know what kind of web compat impact that would have
13:32
<karlcow>
If the
13:32
<karlcow>
required action involves a subsequent HTTP request, it MAY be carried
13:32
<karlcow>
out by the user agent without interaction with the user if and only
13:32
<karlcow>
if the method used in the second request is known to be "safe", as
13:32
<karlcow>
defined in Section 5.2.1.
13:32
<karlcow>
The GET, HEAD, OPTIONS, and TRACE request methods are defined to be safe.
13:32
<karlcow>
DELETE is not safe
13:32
<darobin>
zcorpan: presumably none given that no two browsers do the same thing
13:33
<zcorpan>
darobin: all browsers allow following the redirect in some way (opera/safari get the method wrong, firefox prompts first)
13:34
<zcorpan>
(dunno about ie)
13:34
<darobin>
zcorpan: right, but I doubt anyone relies on this — though I wouldn't be shocked otherwise
13:34
<darobin>
have been too lazy to test IE, might fire up the VM later
13:34
<karlcow>
so basically for XHR it would mean a piece of UI for asking the user which I guess would make some designers go ballistic about the ugly modal window, but that's another story
13:35
<darobin>
karlcow: FF prompts on XHR — it's damned stupid
13:35
<karlcow>
darobin: for all methods
13:35
<karlcow>
or only unsafe methods
13:35
<darobin>
only for unsafe
13:35
<karlcow>
so I'm glad it does
13:35
<darobin>
well, I only tested DELETE here but I know it doesn't prompt on GET redirect
13:36
<karlcow>
There could be nasty side effects.
13:36
<darobin>
karlcow: except that that requirement is built on the assumption that the UA's user knows they're issuing a DELETE over HTTP — that's not the case here....
13:36
<karlcow>
hmm?
13:36
<darobin>
I mean you click a little cross in a UI and you get a modal saying "This site is redirecting you, do you really want that?"
13:37
<darobin>
that's a guaranteed user "WTF???"
13:37
<karlcow>
So you mean the prompt is dumb.
13:37
<karlcow>
Not the fact that there is a prompt
13:37
<darobin>
there is no generic prompt that could make sense here
13:37
<zcorpan>
darobin: or worse, it happens at some random time and the user has no idea what's going on
13:37
<karlcow>
"the prompt message" I meant
13:37
<karlcow>
UX issue
13:37
<darobin>
zcorpan: well yeah, it could happen several ms later, after the UI has removed what's being deleted...
13:38
<zcorpan>
darobin: i mean the author could issue an XHR at any time without it being bound to something the user does
13:38
<darobin>
for XHR, given we're not in an interactive environment, I would expect either the script to have an option to say it knows what it's doing, or to just assume that by default
13:38
<darobin>
zcorpan: indeed, though that's unlikely for DELETE
13:38
<darobin>
but yes
13:39
<zcorpan>
this is the web we're talking about, nothing is unlikely :-P
13:39
<darobin>
heh, sure :)
13:39
<karlcow>
I guess there are two ways to deal with that. Forbid in browsers any unsafe redirections altogether. Always.
13:39
<karlcow>
Or have prose explaining that Browsers should be really explicit about the damages it might create.
13:40
<karlcow>
darobin: XHR, trusting the script is not a good idea.
13:40
<darobin>
karlcow: I would agree with you for stuff like <a href=foo.html method=DELETE> but in XHR I don't think interactive rules should apply
13:40
<karlcow>
think about POST which is unsafe too
13:41
<zcorpan>
this discussions seems moot unless someone's gonna write a spec :-)
13:42
<karlcow>
darobin: hmm… I think I still apply. I do not want to have a script kiddie doing something for me which is not safe.
13:42
<darobin>
I'll mull over making this defined for XHR — fixing HTTP is probably hopeless
13:43
<darobin>
karlcow: I'm not sure I can think of a useful attack scenario that 1) isn't blocked by SOP or 2) isn't already an attack anyway
13:43
karlcow
is looking for SOP acronym
13:43
<darobin>
same-origin policy
13:44
<darobin>
in any case I stumbled over this issue while coding for a project; writing a spec is unlikely to fix the issue within the project's time frame ;)
13:44
<karlcow>
heh
13:54
karlcow
wonders if IE and Chrome still support this https://htaccess.wordpress.com/2009/09/22/x-content-type-options-nosniff-header/
13:54
<karlcow>
X-Content-Type-Options: nosniff
14:06
<darobin>
karlcow: wish I could send that header to my own nose
14:07
<karlcow>
:)
16:20
<TabAtkins>
zcorpan: Now it does, yes, because I fixed it. ^_^ We're now just waiting for someone to write the Color 4 module and actually specify 4/8 digit hex colors.
17:28
<rniwa>
a silly question. is a boolean attribute still considered true even if the value doesn't match its name?
17:28
<rniwa>
e.g. checked=true
17:29
<TabAtkins>
Yes. Booleans are true if they exist, regardless of value, and false if they dont'.
17:30
<TabAtkins>
Ie they use hasAttribute for true/falseness.
17:30
<rniwa>
TabAtkins: apparently the U.S. government doesn't know how to use them correctly
17:30
<rniwa>
:(
17:30
<TabAtkins>
Lots of people don't.
17:30
<rniwa>
TabAtkins: go to https://egov.uscis.gov/crisgwi/go?action=coa.Terms
17:30
<rniwa>
TabAtkins: Accept terms
17:30
<rniwa>
and click on "Yes, this change of address is not for a US Citizen"
17:30
<rniwa>
Ugh… "No,..."
17:31
<rniwa>
then all text fields that are NOT required, has required=true :(
17:31
<rniwa>
and Chrome thinks they're required and won't let me submit the form
17:31
<miketaylr>
i've tried to contact them for about 2 years to fix that
17:31
<rniwa>
miketaylr: maybe we need to fix the browser UI to ignore required fields in some cases?
17:32
<miketaylr>
rniwa: http://my.opera.com/miketaylr/blog/u-s-citizenship-and-immigration-services-change-of-address-bless-their-hearts
17:32
<TabAtkins>
rniwa: That's what your inspector is for.
17:32
<rniwa>
I mean… as ridiculous as it sounds, having to modify the page on Inspector just to omit my middle name, etc… was insanely annoying to say the least.
17:32
<rniwa>
TabAtkins: that's not an acceptable option for a lot of people.
17:32
<TabAtkins>
But required=true is just nonsense. If it was required=false and they were expecting it to not be required, that's a different story.
17:32
<rniwa>
TabAtkins: you can't expect an average user to be able to use this form.
17:33
<rniwa>
TabAtkins: they have "required=no"
17:33
<TabAtkins>
rniwa: No, I can't. They have a broken page.
17:33
<TabAtkins>
rniwa: Oh, you said required=true.
17:33
<rniwa>
TabAtkins: i mean… required=true is also bad.
17:33
<rniwa>
TabAtkins: required=true and required=no are the same thing
17:33
<TabAtkins>
Oh goddammit I just crashed chrome.
17:33
<rniwa>
TabAtkins: as far as HTML is concerned
17:34
<rniwa>
I guess I'm gonna just file a chromium bug about this :/
17:34
<TabAtkins>
Yes, I know. That's what I said earlier. ^_^
17:34
<miketaylr>
they *really* dont' want to fix that page
17:34
<TabAtkins>
Dude, I will shut down that bug. We should *not* go down the rabbit's path of trying to guess what retarded attribute values people try to put in.
17:34
<miketaylr>
"We cannot support Google Chrome for completing your Change of Address at this time. Change of Address is optimized to use Internet Explorer 6 and 7."
17:35
<TabAtkins>
That leads to supporting "false", "n", "no", "off", and more.
17:35
<rniwa>
:(
17:35
<rniwa>
TabAtkins: that's a terrible user experience.
17:35
<TabAtkins>
rniwa: I... agree? That's their fault.
17:35
<rniwa>
TabAtkins: so you're telling me that the correct way to deal with this is to abandon Chrome
17:35
<rniwa>
TabAtkins: and use IE?
17:36
<rniwa>
TabAtkins: I mean that's the only option for many users here.
17:36
<TabAtkins>
That's the right thing *for this page*, because they're complete fucking morons and don't know how to internet.
17:36
<rniwa>
TabAtkins: that's not an acceptable decision as far as I'm concerned.
17:37
<rniwa>
TabAtkins: all non-US citizens living in the U.S. needs to use this change.
17:37
<rniwa>
TabAtkins: that's in the order of millions
17:37
<TabAtkins>
rniwa: I don't think breaking our browser against the spec is the right answer to every single retardely-authored webpage.
17:38
<rniwa>
TabAtkins: what if this was youtube or yahoo.com?
17:38
<TabAtkins>
It's not even broken. It just means you ahve to fill in every field.
17:38
<rniwa>
TabAtkins: would you still say the same?
17:38
<TabAtkins>
It's not.
17:38
<rniwa>
anyway, i'm gonna stop talking about this with you
17:38
<rniwa>
it's not productive
17:38
<TabAtkins>
And if it were, it wouldn't be "optimized to use INternet Explorer 6 and 7"
17:38
<TabAtkins>
Hokay.
21:27
<annevk>
is licensing gonna be clarified for webplatform.org?
21:27
<annevk>
or am I not looking in the right place?
21:27
<annevk>
paul_irish: ^^
21:29
<hober>
annevk: in "Half-day tasks" on http://docs.webplatform.org/wiki/WPD:Getting_Started it says "Content original to the site is licensed under a Creative Commons Attribution License (CC-BY 3.0) for maximum reusability."
21:29
<hober>
(It was on display in the bottom of a locked filing cabinet stuck in a disused lavatory with a sign on the door saying 'Beware of the Leopard'.)
21:30
<annevk>
heh
21:30
<annevk>
maximum would be public domain imo, but I guess that license works
21:31
<hober>
also, wat: http://docs.webplatform.org/wiki/html/elements/custom
21:32
<annevk>
well it originates from Microsoft
21:32
<Hixie>
when they say "original to the site" i presume they're excluding all of msdn which is all i came across
21:32
<Hixie>
(pretty bogus content, too)
21:33
<annevk>
hmm yeah, e.g. http://docs.webplatform.org/wiki/css/color/CSSOM_view
21:33
<annevk>
clearly alpha
21:33
<paul_irish>
yup
21:33
<paul_irish>
we did an automated import of MSDN docs and a lot of items are not exactly standard..
21:34
<paul_irish>
so we need to flag those as proprietary
21:34
<annevk>
what's gonna happen to MDN?
21:34
<paul_irish>
https://hacks.mozilla.org/2012/10/welcoming-the-new-kid-web-platform-docs/ answers that best
21:35
<paul_irish>
there's a little bit of "wait and see" approach from some folks, others in moz want all future docs at WPD
21:35
<paul_irish>
so its a little complicated
21:35
<paul_irish>
but not bad in any way
21:35
<annevk>
ah shit, Mozilla did licensing wrong
21:36
<Hixie>
the agreement when you get an account is kinda weird, too. refers to "Contributions" with a capital C but doesn't define what those are.
21:36
<annevk>
thanks paul_irish!
21:37
<paul_irish>
yeah drop me a link on anything that looks totally whack or shouldnt belong and i'll file a ticket for people to look at it.
21:38
<Hixie>
http://docs.webplatform.org/wiki/html/elements/html_comment_data-type is basically entirely wrong at every level. :-)
21:50
<zewt>
can progressive jpegs please be removed from all jpeg compressors already
21:51
<zewt>
we're not on 300 baud modems and people keep using it by accident and they make my eyes bleed
21:51
<Hixie>
ew no
21:51
<Hixie>
more progressive please
21:51
<zewt>
"progressive jpeg" doesn't mean "progressive decompression"
21:51
<Hixie>
i wish every pic would start off as one pixel of colour and then 4 and then 16 and then 64 etc
21:51
<Hixie>
oh
21:51
<Hixie>
then nevermind
21:52
<zewt>
it's analogous to "interlaced" pngs
21:52
<zewt>
where you get a big hideous blurry blob first
21:52
<Hixie>
isn't that what i just said i wanted?
21:53
<zewt>
dunno, progressive decompression is just "don't wait for the whole image to decompress before showing it", which doesn't require that sort of ugliness
21:53
<TabAtkins>
Hixie: I'd be interested to see if delta-compressing *that* much would be worthwhile.
21:53
<Hixie>
i don't want the images to decompress like window shades
21:53
<Hixie>
TabAtkins: oh i'm sure it'd be negative compression ratio :-)
21:53
<zewt>
blurry images hurt my eyes
21:53
<Hixie>
TabAtkins: at least to do it actually as silly as i describe
21:53
<Hixie>
zewt: just seeing eyes is even worse
21:54
<zewt>
if it starts at a non-blurry resolution and goes up from there, maybe okay (eg. upgrading from regular resolution to retina, so to speak)
21:54
<Hixie>
"regular" is blurry
21:54
<zewt>
but progressive jpegs start as a big artifacty gross mess
21:54
<zewt>
(usually)
22:01
<jgraham>
paul_irish: Repeating what I already said to karlcow and miketaylr, I think at this point the actual content is so confusing that you should add a big "under construction" gif and (more seriously) add links to the "how we want the site structure to end up" pages to the front page
22:02
<jgraham>
It's not really possible to figure out how to help yet because all there is is a big mess of stuff an no obvious coherent structure that it is supposed to be tidied into
22:20
<paul_irish>
jgraham: haha so i totally lost the argument that we should put http://codepen.io/joshnh/pen/ohbHl on the site 5 days ago :)
22:20
<paul_irish>
jgraham: agreed about the IA. thanks for the feedback
22:21
<paul_irish>
ill start some convos on that
22:26
<say2joe>
@paul_irish … love the animated gif
22:27
<say2joe>
(rather the animated gif turned css)
22:35
<TabAtkins>
paul_irish: Out of curiosity, what was the point of doing the under-construction thing with CSS rather than a GIF?
22:35
<paul_irish>
"because", i think..
22:36
<paul_irish>
turns out its smaller.. but only after gzip.
22:36
<TabAtkins>
Sounds legit.
22:36
<TabAtkins>
Likely slower to render, since we don't fast-path solid shadows yet.
22:36
<TabAtkins>
At least, in Chrome.
22:46
<Hixie>
http://codepen.io/joshnh/pen/ohbHl is seriously messed up
22:47
<say2joe>
@Hixie: too much coffee?
22:48
<annevk>
Hixie: using box-shadow or radial gradients via background to draw images is a thing now
22:49
<annevk>
Hixie: hopefully at some point the drawing primitives from SVG just move over so it gets less clunky
22:49
<annevk>
and then we can drop SVG :p
22:49
<annevk>
-> sleep
22:50
<Hixie>
i hope it'll go the way of using html tables for bitmaps...
23:20
<Hixie>
anyone know how many files are in web200904 ?
23:24
<Hixie>
about 600,000 apparently
23:25
<TabAtkins>
I think just count the number of nulls and divide by 3.
23:25
<Hixie>
i wonder what they've been up to since 2009
23:28
<TabAtkins>
Crawling more internet, presumably.
23:28
<Hixie>
you'd think they'd have posted an update in the meantime