02:37
<webesayz>
Hello
02:39
<amites1>
anyone here work with the python version of html5lib?? I'm working on debugging an error with processSpaceCharacters and would appreciate any advice
02:40
<Lachy>
amites1, jgraham and annevk do, but they're not here at this time of night.
02:41
<Lachy>
amites1, gsnedders|work or Philip` too may be able to help though
02:43
<amites1>
doesn't seem like any of them are on right now
02:44
<amites1>
I appreciate your looking out
02:44
<Lachy>
amites1, they'll most likely be here within about 5 or 6 hours when they wake up
02:45
<amites1>
*Grin* if I don't have it figured out by then
02:45
<amites1>
changed servers and upgraded a couple things, everything but PDF generation went smooth
03:46
takkaria
giggles a bit at the WAI procedure thread
04:37
<Hixie>
does python have the equivalent of the C-like ?: operator?
04:54
<gavin_>
Hixie: http://marc-abramowitz.com/archives/2008/05/18/python-ternary-operator/
04:59
<Hixie>
thanks!
08:28
<Lachy>
I can't see what exactly in the proposed procedure that we're not doing, beyond the fact that we haven't yet reached a mutually agreeable solution
08:29
<Lachy>
and that won't happen as long as the WAI PFWG continues to do nothing but push for one specific solution, without accepting or working with any alternatives
09:28
<gsnedders|work>
Lachy: I'm probably not a good person to speak to about html5lib for future reference
10:30
<roc>
I can't stand everyone in public_html being politely dishonest to one another
10:46
<Hixie>
i'm trying to be politely honest
10:50
<MikeSmith>
Hixie: sometimes it's not possible to be both polite and honest at the same time
10:50
<MikeSmith>
and you need to make a choice
10:51
<Hixie>
i usually shut up in those cases
10:51
<MikeSmith>
me too
11:34
<othermaciej>
roc: I'm considering being honestly impolite
11:34
<othermaciej>
but it seems like it would be a waste of time and energy in this case
11:39
<MikeSmith>
othermaciej: I recommend you going with your instincts there.
11:39
<othermaciej>
MikeSmith: sometimes being honestly impolite is useful
11:40
<roc>
it doesn't work too well online unfortunately
11:42
<Lachy>
othermaciej, if you were to be honestly impolite on the mailing list, what would you say?
11:42
<Lachy>
or rather, what point would you want to make?
11:43
<othermaciej>
Lachy, given the people who love to read our logs and send creepy emails about them, that would amount to a cowardly way to post to the list
11:43
<Philip`>
You could say it in our secret IRC channel instead
11:43
<Lachy>
--> #whatwg-cabal
11:44
<othermaciej>
Philip`: how do you know I haven't?
11:44
<Lachy>
shh, don't tell anyone!
11:47
<Hixie>
if you guys do set up a secret cabal list could you please log it so that i can find your discussions when i do google searches for relevant discussion?
11:49
<othermaciej>
can't you use Google's abusive monopolistic control of the internet to just log all the traffic anyway?
11:51
<Philip`>
We could set up the list on Google Wave, so Hixie can use the secret wiretap backdoor to track and analyse the discussions
11:54
<Lachy>
we need a way to allow robots to index the content, but to keep it secret, a way to block unwanted humans from heading it.
11:54
<Lachy>
and we don't yet have a Human Exclusions Protocol (humans.txt)
11:54
<Hixie>
i'm going to go to bed before your conspiracy theories get me in trouble
11:54
<Hixie>
nn :-)
11:57
<MikeSmith>
Lachy: you need to run with that Human Exclusion Protocol idea. do an IETF Internet draft
11:59
<Lachy>
MikeSmith, I was just thinking about an April Fools RFC for that :-)
11:59
<dbaron>
Would it apply if Koko the Gorilla starts using a Web browser?
11:59
<MikeSmith>
heh
12:01
<Lachy>
Human-Agent: Mike*
12:01
<Lachy>
Disallow: /
12:01
<gsnedders|work>
MikeSmith: Quick, change to billyjackass!
12:11
<othermaciej>
hmm, reading some blog posts for the past few weeks it seems like a remarkable number of people really had faith in XHMTML2
12:19
<mookid>
is it going to be possible for javascript to render other formats like PDF in html5?
12:20
<gsnedders|work>
mookid: I guess the main issue with that is lack of a binary data type
12:20
<mookid>
hmm
12:20
<mookid>
ok
12:20
<mookid>
frustrating because the only option right now is to popup a new window
12:20
<gsnedders|work>
mookid: Otherwise, I don't think there's anything stopping you (from converting to either HTML/SVG elements, or rendering onto a canvas)
12:20
<mookid>
eww that's pretty gnarly
12:20
<gsnedders|work>
mookid: Not just use object or something?
12:21
<gsnedders|work>
mookid: Well, how else would you render it with Javascript?
12:21
<gsnedders|work>
mookid: Am I missing what you meant?
12:21
<mookid>
situation I'm in - if I make XMLHttpRequest to a pdf document
12:21
<mookid>
how do I get that rendered?
12:22
<gsnedders|work>
Convert it to a data URI and set an object's data to it?
12:22
<gsnedders|work>
Why not just set the object's data to the URL you request with XHR?
12:23
<mookid>
cos it's got custom headers :P
12:23
<gsnedders|work>
I was guessing that
12:23
<mookid>
there is one way that I guess would work - which is to render a flash app and pass the data into that
12:23
<gsnedders|work>
If it contains any null bytes, be warned that Opera at least truncates responseText at that
12:24
<gsnedders|work>
Converting it to a data URI should work, though
12:24
<mookid>
yeah XMLHttpRequest is pretty bad
12:24
<mookid>
are there any plans to update/fix that?
12:24
<mookid>
can we at least drop the XML infront of it?
12:25
<gsnedders|work>
The null issue? I guess the answer is I can't comment on that.
12:25
<gsnedders|work>
We can't really drop the XML in the name because thousands of sites rely upon it being there
12:25
<gsnedders|work>
Http is equally misleading, as you can make FTP requests too in theory
12:25
<gsnedders|work>
(or any other scheme)
12:25
<mookid>
can't you just change it and make the old name a pointer?
12:25
<gsnedders|work>
Yes, but is it worthwhile changing that?
12:26
<mookid>
well I think it can be improved anyway
12:26
<mookid>
the spec could be more vigorous wrt the HTTP spec
12:27
<gsnedders|work>
We're gonna have to keep backwards compat with XHR forever, more or less, so adding an alias doesn't help.
12:27
<gsnedders|work>
Vigorous in what way?
12:27
<mookid>
3xx forwards are very bad right now in XHR
12:27
<mookid>
+ the object could be structured in a more intuitive fashion
12:28
<MikeSmith>
http://cloudofdata.com/2009/07/does-linked-data-need-rdf/
12:28
<mookid>
nah it doesnt
12:28
<mookid>
:P
12:28
<gsnedders|work>
mookid: That can't really be changed for backwards compat. reasons
12:28
<mookid>
gsnedders|work: ok so is there room to create a new client library?
12:29
<gsnedders|work>
mookid: I doubt it'll be added at a browser level (on grounds that XHR already exists)
12:30
<gsnedders|work>
mookid: What's your issue with redirects?
12:30
<mookid>
well they're inconsistent and the object makes it very hard to get a handle on what's actually going on
12:30
<mookid>
the implementations are inconsistent
12:31
<gsnedders|work>
That's an implementation issue, and all are moving towards the current WD
12:31
<gsnedders|work>
Actually, I expect the current ED
12:32
<mookid>
maybe I'm not up enough on XHR but redirects follows are automated for some 3xx codes right?
12:32
<gsnedders|work>
For "HTTP redirects"
12:36
<mookid>
scariest idea ever
12:37
<mookid>
give your javascript sandbox namespaces a URI
12:37
<mookid>
then I can give an object a data URI
12:39
<mookid>
so essentially what you're saying is I can't use XHR response data to render a PDF?
12:39
<gsnedders|work>
huh?
12:40
<mookid>
nvm =)
12:40
<mookid>
having an accept header attribute would make my life a lot easier.. ;D
12:41
<gsnedders|work>
Take the responseText, encode as base64, then add before that data:application/pdf;base64,
12:41
<gsnedders|work>
How does that not work?
12:41
<mookid>
what do you do with that from there though?
12:42
<mookid>
I can't render it or make it available to save to disk in a browser sandbox
12:42
<gsnedders|work>
var obj = document.createElement("object"); obj.data = foobar;
12:42
<gsnedders|work>
Then add the obj and if there is a PDF plugin installed, it'll render.
12:42
<gsnedders|work>
*add the obj to the tree somewhere
12:42
<mookid>
oh right hmm lol
12:43
<mookid>
I'm sure I read somewhere you can't do that
12:43
<mookid>
is that new?
12:44
<gsnedders|work>
No, but I'm pretty certain that should work
12:44
<gsnedders|work>
If it doesn't, you've hit browser bugs
12:44
<mookid>
sound, cheers
12:44
<gsnedders|work>
There may be some other way that I haven't thought off, though.
12:44
<gsnedders|work>
Do PDFs contain null bytes often?
12:45
<mookid>
haven't looked into that yet
12:45
<gsnedders|work>
As I say, any containing null bytes will fail in Opera
12:46
<mookid>
ok bear with me here - if the browser has no plugin installed - there's no way to provide a save to disk dialogue box?
12:51
<mookid>
MikeSmith: was that link for me?
12:54
<gsnedders|work>
mookid: What the browser does in that case is implementation dependant behaviour.
12:54
<gsnedders|work>
mookid: But there's no way to force a browser to provide a save dialogue without running in a privileged state
13:13
<MikeSmith>
mookid: what link?
13:13
<mookid>
the RDF one
13:14
<mookid>
RDF vs. LinkedData
13:14
<MikeSmith>
nope
13:14
<mookid>
k
13:14
<MikeSmith>
that was for all us, for our collective edification
13:14
<mookid>
sweet
13:14
<MikeSmith>
mookid: you have an interest in linked data?
13:14
<mookid>
Hmmm - I'd like to say yes
13:14
<mookid>
but honestly I'm not realy qualified to say that
13:15
<mookid>
I know REST and I'm working on a layered platform
13:15
<mookid>
but I'm not using RDF
13:15
<mookid>
so that article sounds good to me
13:15
<mookid>
granted.. RDF seems like a pretty nice tool for 'fractal' data graphs
13:16
<mookid>
it's just more cost to my project right now
13:16
<mookid>
I'm all set up to introduce an additional RDF layer whenever I need :)
13:16
<mookid>
thanks to URI's and content negotiation
13:16
<MikeSmith>
sometimes it costs more to add stuff later
13:17
<MikeSmith>
I don't actually know much about RDF beyond the very basics
13:17
gsnedders|work
still isn't sure using URIs to identify things is entirely sensible
13:17
<gsnedders|work>
I mean, what URI do I use for my grandmother?
13:17
<mookid>
...
13:18
<mookid>
there's a joke there somwhere
13:18
<MikeSmith>
btw, a note while I'm thinking about it: If somebody ever does do XML5, it should relax the restriction that comments can't contain "--"
13:18
gsnedders|work
still thinks XML5 should be a strict subset of XML 1.0 4th edition
13:18
<MikeSmith>
along with all other "backwards compatible with SGML" constraints
13:19
<mookid>
gsnedders|work: it's all to do with how you actually define a Resource
13:19
MikeSmith
is reminded that he meant to open a bug about the phrase "backwards compatible" in the spec intro
13:20
<mookid>
MikeSmith: if I wanted to introduce RDF to my system I'd only have to add another layer of RDF representations ontop of my existing URIs
13:21
<mookid>
and I can do that easily with Accept/Content-Type headers
13:21
<Philip`>
gsnedders|work: What do you mean by "strict subset"? (All well-formed XML1 documents must parse to the same DOM when using the XML5 algorithm? That sounds more like a superset to me)
13:21
<MikeSmith>
can a Resource just be a particular stream of data transmitted over TCP/IP?
13:21
<gsnedders|work>
Philip`: I mean in terms of document conformance.
13:22
<gsnedders|work>
Philip`: But yes, the parsing should be a superset, and conforming a subset
13:22
<Philip`>
gsnedders|work: So all conforming XML5 documents should also be conforming XML1 documents, but not the other way around?
13:22
<gsnedders|work>
Philip`: Yeah.
13:22
<gsnedders|work>
Philip`: Because DOCTYPE is evil.
13:22
<Philip`>
gsnedders|work: Okay, makes sense
13:23
<mookid>
gsnedders|work: why do you say that?
13:23
<Philip`>
I suppose my main complaint with XML is not that draconian error handling is bad, but that it's too complex to follow the rules in practice
13:24
<mookid>
MikeSmith: technically, yes - if you were logging TCP/IP traffic you could give each packet a URI
13:24
<mookid>
but the actual data trasmitted is a representation
13:25
<Philip`>
e.g. properly-nested tags and quoted attributes are pretty easy, and the pain is in the complex character requirements and the forbidden sequences like "]]>" and "--"
13:25
<Philip`>
(and in doctypes)
13:26
<mookid>
could you use JSON?
13:26
gsnedders|work
notes XHTML, Atom, and RSS are not JSON, so for most people using XML the answer is no.\
13:27
<mookid>
I mean use JSON for HTML5
13:27
<gsnedders|work>
If you define a JSON serialization of a DOM, sure.
13:27
<mookid>
well isn't it better suited than XML?~
13:28
<mookid>
I thought there's an issue with XML in terms of what you can actually assume about the data from the order/tag heirarchy
13:29
<mookid>
there must be most of that work already done within apps like firebug..?
13:30
<Philip`>
Someone should modify something like http://canvex.lazyilluminati.com/misc/sexp.html to work with JSON
13:30
<mookid>
:)
13:30
<mookid>
Kris Zyp is/was working on JSON-Schema and json references
13:31
<mookid>
where schema's all linked together and could reference one another /themselves
13:32
<mookid>
I dunno exctly how fruity the DOM is but I would imagine you can break it down quite neatly
13:37
<gsnedders|work>
Anyone from Mozilla around?
13:37
<gsnedders|work>
Or anyone with a clue of how to report a bug in Gecko?
13:37
<gsnedders|work>
https://bugzilla.mozilla.org/page.cgi?id=broken-website.html says to go through the Help menu, but I don't have a "Report Broken Web Site" in the nightly I have.
13:38
<Philip`>
gsnedders|work: Just use Bugzilla normally, and file a bug with a test case?
13:38
<gsnedders|work>
Philip`: You aren't allowed to do that nowadays.
13:38
<gsnedders|work>
Oh well, their problem.
13:38
<Philip`>
Uh, not allowed to file bugs?
13:39
<gsnedders|work>
"IMPORTANT: if the problem is with a broken web page, you need to report it a different way."
13:39
<gsnedders|work>
So seeming you're not.
13:40
<Philip`>
That sounds like it's meant for bug reports like "I visited Yahoo Mail and it doesn't work"
13:41
gsnedders|work
grumbles
13:41
<Philip`>
and I'd hope you have more detail and a minimised test case and then you could file a proper bug :-)
13:41
<gsnedders|work>
I'm reporting one of their tests is broken (and so is their behaviour that passes it)
13:44
<Philip`>
gsnedders|work: Then just fill out the bug form and put all the relevant details in it
13:47
gsnedders|work
finds through one really obscure bit of the spec that the tests are right, and he is wrong.
17:04
<hallvors>
why doesn't the spec allow multi-letter accesskey values?
21:28
<Darxus>
I'm having a surprisingly difficult time finding XHTML 1.0 and HTML 5 test suites.
21:31
<takkaria>
what would an XHTML1 test suite test?
21:31
<jcranmer>
HTML 3.2, obviously
21:32
<takkaria>
I mean, I wasn't aware there were many testable things in XHTML1
21:39
<Philip`>
takkaria: I assume it normatively includes HTML4, so you could test all of HTML4's conformance requirements
21:39
<takkaria>
I was just trying to find thoser but it looks like w3.org is having problems
21:44
<Darxus>
takkaria: Browser conformance.
21:53
<Darxus>
There seems to be no good statistics on how well browsers support XHTML 1.0, and as that is the last standard to support text/html, and IE doesn't support application/xhtml+xml, and it was released 9 years ago, I think that it qualifies as the most important current standard. And I think there should be better browser support statistics on current standards.
21:54
<jcranmer>
XHTML is predominantly treated as HTML 4.0-serialized-as-XML
21:54
<jcranmer>
(from my experience)
21:56
<Darxus>
jcranmer: So you would expect browser conformance to be the same between HTML4 and XHTML 1.0?
22:00
<Lachy>
Darxus, XHTML as text/html is, as far as the browser is concerned, simply HTML.
22:01
<Darxus>
Lachy: Thank you.