00:26
<MikeSmith>
cardona507: hey
00:28
<cardona507>
howdy MikeSmith- how r u?
00:32
<MikeSmith>
doing good
05:39
<Hixie>
someone keeps hammering whatwg.org with a bot that (a) has a bogus User-Agent field and (b) is idiotically implemented
06:21
<MikeSmith>
Hixie: what is the bot requesting?
06:22
MikeSmith
hopes it's not some script of his gone awry
06:28
<annevk5>
hmm, is that the same bot that was hammering html5.org at some point?
06:48
<Hixie>
MikeSmith: the spec, but they're fetching every fragment identifier one after the other as a separate doc
06:53
<annevk5>
http://wearehugh.com/public/2010/08/html5-video-accessibility/ heh
06:58
<Hixie>
anyway i solved the problem by just having a script monitor my server and if it sees anyone doing that, it bans their IP
07:03
<annevk5>
Hixie, "registrying"
07:04
<Hixie>
thanks fixed
07:05
<annevk5>
it seems you didn't add the reference WEBLINK
07:06
<Hixie>
it was already there
07:07
<Hixie>
(my build process won't let me forget a reference, it aborts before doing the regen if there are any broken ones)
07:11
<annevk5>
hmm -- http://gregbaker.ca/blog/2010/08/07/p-n-np/
07:16
<Hixie>
i doubt the guy was happy with that
07:17
<annevk5>
it would have been more fun if it was the other way around, yes :)
07:21
<Hixie>
no i mean happy that people put it in the public web
07:21
<Hixie>
i think at this point people would have been dumbfounded if someone found P=NP
07:45
<roc>
P=NP would immediately trigger the singularity
07:45
<xfinx>
mogguh
08:12
<MikeSmith>
hmm, it seems Chrome now intentionally disallows XHR with relative URLs in scripts running locally
08:12
<MikeSmith>
http://code.google.com/p/chromium/issues/detail?id=37586
08:14
<MikeSmith>
http://code.google.com/p/chromium/issues/detail?id=37586#c4
08:14
<MikeSmith>
"In order to be more secure (and indeed now the most secure of any browser as you note ;-) Chromium now restricts a local HTML file from reading other local files."
08:16
<MikeSmith>
so you can't test XHR apps locally in Chrome if you're using open() with a relative URL
08:24
<everton>
what's up? ^^
08:26
<MikeSmith>
http://code.google.com/p/chromium/issues/detail?id=47416
08:27
<MikeSmith>
heh http://code.google.com/p/chromium/issues/detail?id=47416#c1
08:28
<MikeSmith>
"I know we've had a few vocal complaints, but the volume feels low relative to our millions of users userbase. … do we have a way of quantifying whether people are annoyed enough about a given thing to warrant a change? … I've only seen comments on the bugs and nothing like blogs or news stories or end user complaints in the Chrome user forums."
08:31
<annevk5>
set up a local Apache server and it works fine :)
08:31
<MikeSmith>
super
08:31
<annevk5>
testing over file URLs is dangerous anyway because the API works quite different
08:31
<everton>
unforgetable
08:32
<annevk5>
e.g. send() throws rather than returns a 404 for stuff that cannot be found
08:32
<MikeSmith>
annevk: hmm, OK, didn't know that
08:32
<annevk5>
well, your file system is not HTTP :)
08:32
<MikeSmith>
yeah, sure
08:32
<annevk5>
imo we should only support ftp/http/https for XHR and extend it if someone provides a use case
08:33
<annevk5>
rather than the somewhat theoretical open ended model that is defined now
08:33
<MikeSmith>
seems like that would be fine if all browsers changed to have that same behavior
08:34
<MikeSmith>
as it stands now, Chrome is the only one that has this restriction
08:34
<MikeSmith>
and the error message it emits for this case is suboptimal
08:36
<MikeSmith>
anyway, dude has a really great strategy there for evaluating how much a particular browser-behavior change is going to screw things up for Web developers: Just make the change, then wait for news stories and blog posts from people pointing out their apps don't work any more when they test them locally
09:21
<annevk5>
is there some kind of "is same Document" method?
09:22
<Workshiva>
Compare the documents?
09:22
<annevk5>
thought it might be fun to have the browser serialize send() stuff the server echoing it back and then the browser parsing again and seeing if stuff is identical
09:22
<annevk5>
aah
09:22
<annevk5>
isSameNode()
09:23
<annevk5>
oh no
09:23
<Workshiva>
I assume you're looking for structural similarity, not object identity
09:23
<annevk5>
isEqualNode()
10:03
<annevk5>
did Firefox change the way it encodes URLs for XMLHttpRequest recently?
10:03
<annevk5>
it used to force UTF-8
10:03
<annevk5>
now it seems consistent with HTML form submission
10:03
<annevk5>
which is wrong per XMLHttpRequest
10:03
<annevk5>
bah
10:04
<annevk5>
hopefully that can still be fixed as it would be nice to keep the URL encoding madness as restricted as possible
10:17
<annevk5>
Hixie, can you look into that XMLHttpRequest email?
10:17
<annevk5>
Hixie, it's blocking a test and I'm feeling impatient ;p
10:18
<annevk5>
hmm, is about:blank same-origin?
10:18
annevk5
checks
10:21
<annevk5>
ah
10:21
<annevk5>
the for URLs section indicates about:blank is non same-origin
11:12
<gsnedders>
whatwg.org down?
11:12
Philip`
finds no evidence to the contrary
11:26
<annevk5>
gsnedders, wfm
11:27
<annevk5>
gsnedders, maybe you're blocked by Hixie's script?
11:27
<gsnedders>
Now wfm
11:28
<annevk5>
these mime advocates...
11:28
<annevk5>
for testing responseText you first need to check the charset parameter; if the type is text/html or XML you need to check per those specs; then check the BOM and finally fallback to UTF-8...
11:28
<annevk5>
just UTF-8 would be so much better
14:29
<MikeSmith>
gsnedders: you around?
14:33
<MikeSmith>
wanted to ask if it'd be possible to add an option to anolis to cause it to add id attributes to any <a> elements in the source that don't already have an id
14:34
MikeSmith
takes a look at the code
14:46
<gsnedders>
MikeSmith: I've really not touched Anolis in ages.
14:46
<MikeSmith>
OK
14:47
<gsnedders>
Apparently wearing makeup will help me get into pubs where you have to be 20 to get in…
14:47
<MikeSmith>
like war-paint type of makeup?
14:47
<gsnedders>
I think the word "pink" was mentioned.
14:48
<jgraham>
It was?
14:48
<jgraham>
Anyway, I'm not sure the drag queen look is a winner
14:48
<gsnedders>
Nor am I.
14:48
<cheeser>
does it get you free drinks inside?
14:48
<cheeser>
'cause then it might be... ;)
14:49
<gsnedders>
cheeser: Not relevant if you don't get in.
14:49
<cheeser>
true.
14:49
<cheeser>
though i find such places not worth visiting anyway
14:50
<cheeser>
i just want to drink not make a statement about my social worth.
17:48
<AryehGregor>
whatwg.org slow?
17:50
<AryehGregor>
More like not working at all (the spec specifically).
17:50
<AryehGregor>
Chrome automatically suggests the equivalent section of the W3C spec.
17:50
<AryehGregor>
Sneaky Google.
17:51
<AryehGregor>
It got the WD, though, not the ED.
17:57
<jgraham>
AryehGregor: Everything Hixie related is dead I think
17:57
<AryehGregor>
oh noes :(
17:58
<AryehGregor>
SSH works.
17:58
<Hixie>
all seems to work for me...
17:58
<Hixie>
just a bit slow
17:58
<Hixie>
someone at opera is hammering the server
17:58
<Hixie>
something to do with the selectors test suite
18:01
jgraham
blames gsnedders
18:02
<jgraham>
(no evidence it is him)
18:02
<jgraham>
(apart from a general theory about what is going on)
18:03
<gsnedders>
Well, even then it's not my problem…
18:03
<gsnedders>
But all the people whose problem it would ordinarily be have gone home
18:03
gsnedders
sighs
18:05
<gsnedders>
(Provided the theory is right)
18:06
<jgraham>
Hixie: You should put everything on your server into a convenient bundle so we can download it once :)
18:08
<AryehGregor>
I'd appreciate outside perspectives on my argument with Justin Lebar on the whatwg list.
18:10
<TabAtkins>
Which thread is that, Aryeh?
18:11
<AryehGregor>
TabAtkins, the one on resource packages.
18:11
<TabAtkins>
Ah, kk. I agree with you completely.
18:11
<AryehGregor>
Well, then say so on the list. :)
18:12
jgraham
feels like the whole resource packages thing could just lead to a world of pain
18:12
<AryehGregor>
Yeah, it does seem complicated and error-prone.
18:13
<AryehGregor>
But it's about as simple as you can get without requiring server changes . . .
18:13
<jgraham>
and adding complexity like letting the browser choose what to do depending on the cache state or whatever would make it much much worse
18:13
<Philip`>
jgraham: More or less pain than a world without resource packages?
18:13
<AryehGregor>
But there are plausible cases where not letting the browser mix cached copies of things would make resource packages a huge performance *loss*.
18:13
<jgraham>
Philip`: Like a more painful world in which some things are slightly faster
18:13
<Hixie>
AryehGregor: doesn't matter who agrees with whom on the whatwg list
18:14
<Hixie>
AryehGregor: i only consider the arguments
18:14
<Hixie>
e-mails like "i agree" don't even make it to my pile
18:14
<AryehGregor>
Hixie, we're talking about a spec edited by Justin Lebar, not you.
18:14
<Hixie>
ah
18:14
<Hixie>
then what's it doing on the whatwg list? there aren't any other editors there at the moment :-P
18:14
<Hixie>
(only kidding)
18:17
<jgraham>
AryehGregor: If the cases where it can be slower can be predicted by the author in advance it doesn't seem so bad. If the cases don't happen so often it doesn't seem so bad. If the cases are frequent and serious we should probably throw out the whole idea rather than try to add more3 complexity
18:18
<AryehGregor>
jgraham, you'll already see inconsistency if the packaged copy doesn't match the real one, so I don't think it's much more complexity to say the caches can be shared.
18:18
<AryehGregor>
I'd imagine it's less implementation complexity to share the caches.
18:27
<Hixie>
someone e-mailed me asking about the whatwg status annotation password thing (the one that gets sent to you when you log in in the spec)
18:27
<Hixie>
i didn't understand the question they were asking so i asked them to clarify
18:28
<Hixie>
what they're askingme is why the firefox beta is not rendering any pages
18:28
<Hixie>
...
18:28
<Hixie>
. . .
18:29
<kbrosnan>
Hixie: point them at http://support.mozilla.com/en-US/kb/Ask+a+question
18:29
<kbrosnan>
if you have the time
18:30
<Hixie>
the hybi list went from not wanting a variable-width length to wanting a variable-width length on the basis that saving 7 bytes is critical
18:30
<Hixie>
kbrosnan: too late, but thanks
18:30
<TabAtkins>
...
18:30
<TabAtkins>
ffs
18:41
<AryehGregor>
There are a heck of a lot of obsolete elements and attributes.
18:46
<Hixie>
and that list doesn't include obsolete values :-)
19:31
<henrikbjorn>
Hey :p would this https://gist.github.com/076fbf047a23cbfd6c2e be valid as a page footer or is there a better or more correct way ?
19:42
<nimbupani>
henrikbjorn: I think an article would be more appropriate than a section there.
19:42
<henrikbjorn>
its going to container something like "there is now 2323 people online@
19:42
<henrikbjorn>
"
19:42
<henrikbjorn>
now you say it i think your right
19:59
<AryehGregor>
So it turns out that when you get form.action as an IDL attribute when the DOM attribute is "", there's a reason you get "" instead of the URL being resolved. Probably a bad reason, though.
20:00
AryehGregor
files a bug
20:53
<AryehGregor>
In JavaScript, how could I test whether a given exception is a DOMException rather than some other type of exception? Or is there no way to do that?
20:53
<gsnedders>
Given Web DOM Core impl: e instanceof DOMException
20:54
<gsnedders>
I don't think it's guaranteed under DOM Level 3 Core
20:54
<AryehGregor>
It'll do for now. Thanks.
20:57
gsnedders
just searched for WedIDL. Obviously he wants to get married.
20:58
<AryehGregor>
I'm going to a wedding in less than two hours, as it happens.
20:59
<AryehGregor>
Actually, I'm going to three weddings of friends over the course of less than three weeks.
20:59
<AryehGregor>
And there are at least two other people I know who are getting married in the same period, but I'm not going to their weddings.
20:59
<AryehGregor>
Jews are apparently very uncreative in deciding when to get married.
21:01
<Philip`>
Approximately 50% of people in the world are so uncreative that they get married at exactly the same time as someone else, and in exactly the same place
21:02
<AryehGregor>
True. I guess I can't complain.
21:11
AryehGregor
just misspelled WebIDL as WedIDL too
21:33
<AryehGregor>
This gives me an OOM error in the console on Opera . . . o_O http://aryeh.name/tests/reflection.html
21:33
<AryehGregor>
10.60 on Ubuntu 10.04. Can anyone reproduce?
21:33
<AryehGregor>
Kind of inconvenient for testing purposes.
21:34
<AryehGregor>
(and yikes, neither Gecko nor WebKit handles numeric IDL attributes anything like what the spec says)
21:35
<TabAtkins>
Urgh, the difference between IDL and [sg]etAttribute with large values is kinda scary.
21:36
<jgraham>
AryehGregor: I wonder if testing DOMException === e.constructor is supposed to work
21:37
<jgraham>
It feels like it should but I haven't read the relevant specs in detail
21:37
<jgraham>
Works in Gecko, not in Opera and Webkit failed to throw at all in my testcase
21:38
<jgraham>
AryehGregor: I see the OOM message
21:39
<AryehGregor>
el.setAttribute(x, y) followed by el.getAttribute(x) should always return (string)y, right?
21:39
<AryehGregor>
There is no way I'm going to file all these Gecko and WebKit bugs individually, that would be crazy.
21:41
<jgraham>
AryehGregor: You should turn the tests into a giant smiling face remeniscient of early 90s rave culture. Then vendors will be compelled to fix them
21:41
<jgraham>
without you filing any bugs
21:41
<AryehGregor>
Or maybe I'll just submit the tests to the W3C, that might also get their attention (you'd hope).
21:41
<jgraham>
AryehGregor: Yes, please do
21:42
<AryehGregor>
I was planning to iron them out some more.
21:42
<jgraham>
I can help with that if necessary
21:42
<AryehGregor>
You know, go through all the failures and figure out which of the tests are bad.
21:42
<AryehGregor>
But there are rather a lot of failures by now.
21:42
<AryehGregor>
Amazing how everything fell apart when I got to numbers, was doing okay before then.
21:43
AryehGregor
has only filed six bugs so far: http://aryeh.name/tests/bugs_filed.txt
21:44
<jgraham>
Is it me or is your server being slow?
21:46
<AryehGregor>
The page usually doesn't start rendering until the script finishes executing.
21:46
<AryehGregor>
I'm running the tests inline, not onload.
21:47
<jgraham>
AryehGregor: It might be nice to port some of your code to the test harness in the repo so that it is easy to reuse in other contexts
21:47
<AryehGregor>
So I think that's probably the problem.
21:47
<AryehGregor>
What test harness in what repo?
21:47
<jgraham>
AryehGregor: (I meant the git part)
21:47
<AryehGregor>
Oh.
21:47
<jgraham>
AryehGregor: (but I have what I want now so it is fine)
21:47
<AryehGregor>
Fast for me.
21:47
<jgraham>
AryehGregor: In the html5 repo
21:48
<jgraham>
http://dvcs.w3.org/hg/html/file/f1101b5cbb47/tests/resources/testharness.js
21:49
<jgraham>
That is being reused in some other contexts (e.g. bits of WebApps stuff)
21:49
<AryehGregor>
Looks like it would be easy to port.
21:49
<jgraham>
So it would be nice if one could do assert_{something} to test the properties on interfaces reflect in the expected way
22:01
<annevk5>
all XHR tests use that framework
22:01
<annevk5>
and EventSource too
22:01
<annevk5>
hopefully gonna release XHR tomorrow, not sure about EventSource
22:34
<Hixie>
Philip`: ok i can register punycoded subdomains again if you still want a one-character variant of c.whatwg.org to go to multipage... what character do you want me to use? c-cedilla you said wasn't convenient? (it's alt-c on mac en-US, but I don't know what's conveinient in your locale and platform)
22:50
jgraham
wonders what is wrong with "m", really
22:50
<jgraham>
Or are we expecting a mobile version of the site soon?
22:51
<jgraham>
"d" is also easy to type on a variety of keyboard layouts and operating systems
22:54
<Hixie>
"m" is boring
22:55
<aho>
µ
22:55
<jgraham>
Hixie: But functional...
22:55
<Hixie>
pah
22:56
<Hixie>
we have two multipage specs, which should "m" go to?
22:56
<jgraham>
If I were going to be less utilitarian in one part of whatwg.org I would not chose the domain names
22:57
<annevk5>
have "a" go to complete/ (standing for "all")
22:57
<jgraham>
I thought the deal was that ordinary spe was whatwg.org/html5
22:57
<jgraham>
anyway I am also open to mc.whatwg.org
22:57
<jgraham>
for multipage complete
22:58
<annevk5>
but really whatwg.org/C wfm
22:58
<annevk5>
also, XHR and HTML5 email?
22:58
<annevk5>
o_O
22:58
<annevk5>
time to get some sleep
22:58
<annevk5>
jgraham, left a bunch of comments for you in some other channel
22:58
<annevk5>
jgraham, about seven hours ago
22:59
<jgraham>
annevk5: Replied o them, I think
22:59
<jgraham>
*to
22:59
<jgraham>
But I may have missed osme
22:59
<jgraham>
*some
22:59
<annevk5>
oh, guess I'll see tomorrow then
22:59
<annevk5>
my other laptop has the logs
22:59
<jgraham>
I fixed assert_throws a bit anyway
23:00
<jgraham>
If there was anything else I missed it
23:01
<annevk5>
guess i need to resync with that file tomorrow then