00:00
takkaria
wonders how the mobile IE works on the real web
00:01
<hsivonen>
takkaria: what's the relationship to normal IE codebase-wise?
00:01
<Hixie>
on this page:
00:01
<Hixie>
http://www.hixie.ch/tests/adhoc/html/screen-readers/
00:02
<Hixie>
it says "Index of /tests/adhoc/html/screen-readers" three times
00:02
<Hixie>
and says "eighty" twice (at the end of a line each time)
00:02
<Hixie>
why?!?!
00:02
<webben_>
it should say the first thing twice because it is in both title and h1
00:03
<takkaria>
hsivonen: I have no idea. I'm getting a windows mobile smartphone tomorrow, though, so I guess I'll see how well it works then...
00:03
<takkaria>
I guess I can get opera mobile or whatever the portable mozilla browser's called though
00:03
<hsivonen>
http://www.lukew.com/resources/articles/PSactions.asp
00:04
<Hixie>
webben_: twice i would understand
00:04
<takkaria>
Hixie: 80? is that the width of the line or something?
00:04
<Hixie>
(though it's still retarded)
00:04
<Hixie>
takkaria: maybe? but why does it say it, even then?!
00:05
<webben_>
Hixie: What happens in Fox?
00:05
<karlUshi>
nah it is not 80, but "s'ti" in Quebecois. It means something like "geez"
00:06
<Hixie>
webben_: dunno, don't want to go through the pain of switching browsers yet!
00:06
<takkaria>
I think part of the problem of screenreaders for the web is that there isn't actually a web-specific screnreader
00:06
<takkaria>
one of those could do a much better job, I imagine
00:07
<webben_>
takkaria: There are. But there shouldn't be any need for such a thing.
00:07
<webben_>
takkaria: HTML is in a lot of different applications, don't forget, not just web browsers.
00:07
<webben_>
e.g. WebKit is used in Colloquy and Mail
00:08
<takkaria>
mm, true. but in applications it tends to be slightly different, since they're not browsers
00:08
<karlUshi>
plus the fact that navigating the web is not the only thing
00:08
<karlUshi>
you have to be able to use the desktop as large
00:08
<webben_>
and the process of transmuting HTML to a screen reader-navigable DOM is very similar to that for Word files and PDF files and everything
00:09
<webben_>
takkaria: There have been many attempts at providing dedicated browsers. But there are major advantages to using the same software as sighted peers.
00:09
<Hixie>
i tend to agree that there should be a dedicated speech browser rather than a screen reader, but that's a separate problem entirely
00:09
<takkaria>
fair enough. I can see that
00:10
<takkaria>
but if HTML5 works out, then I would imagine that a dedicated browser would be much easier than at present
00:10
<Hixie>
sure wish i knew why it kept saying "ninety eight percent"
00:10
<webben_>
Hixie: That's page load information.
00:10
<webben_>
i.e. how much of the page has loaded
00:11
<webben_>
Hixie: quite possibly that was your 80 earlier too
00:11
<Hixie>
no, IE is not doing anything
00:11
<Hixie>
maybe, though i doubt it, it happens in the same place on that page
00:13
karlUshi
would suggest hixie to learn with a blind user who is using a lot jaws.
00:13
<takkaria>
Hixie: are you IRCing on the same machine as the one you're using JAWS on?
00:13
<karlUshi>
each category of individuals develops skills with regards to software, and there is always a learning curve
00:14
<Hixie>
takkaria: no
00:15
<Hixie>
karlUshi: this is not software that should have a notable learning curve. the target audience includes the learning disabled.
00:15
<Hixie>
i seem to have entered some mouse-related reading mode
00:15
<Hixie>
maybe something to do with the cursor settings
00:15
<karlUshi>
cough cough :) wishful thinking :)
00:16
<webben_>
Hixie: Heh. So does Windows. :)
00:16
<Hixie>
karlUshi: this is orders of magnitude harder than it needs to be
00:16
<Hixie>
i can't work out how to get this out of this strange mode
00:16
Hixie
logs out and back in agaain
00:17
<webben_>
Hixie: Have you played around with Window-Eyes yet?
00:17
<Hixie>
no
00:17
<webben_>
I've tended to find that less overwhelming than JAWS.
00:17
<karlUshi>
how do *you* know what *needs* to be, again look at my previous comment, spend time with a real user :) and identify the real problem. :)
00:17
<karlUshi>
moving to svgopen
00:19
<Hixie>
webben_: ah
00:19
<Hixie>
christ this weird ass mode followed my to the logon box
00:19
<Hixie>
it's only reading the stuff from the disabled dialog
00:19
<webben_>
Hixie: What happens in this mode exactly?
00:20
<Hixie>
the dialog that's in front telling me my password is wrong is not reading it
00:20
<Hixie>
s/it//
00:20
<takkaria>
hsivonen: that forms design post is very interesting, ta
00:20
<Hixie>
webben_: hard to describe
00:21
<Hixie>
so "say all" doesn't work in dialogs, i guess
00:22
<webben_>
I don't think it would
00:22
<Hixie>
i hit "say paragraph" and it says "blank"
00:22
<Hixie>
helpful, thanks jaws
00:22
<webben_>
Hixie: try just tabbing around?
00:22
<webben_>
(if this is a dialog)
00:22
<Hixie>
webben_: an alert dialog, i mean
00:23
<Hixie>
webben_: pressing tab just makes it say "tab".
00:23
<Hixie>
i can't get it to read me the message again
00:23
<webben_>
Hixie: try INSERT + B (read current window)
00:24
<Hixie>
wait, i've found the bug
00:25
<Hixie>
(i'm in laptop mode, btw, so i use caps lock instead of insert)
00:25
<Hixie>
caps lock Y dismisses the alert dialog without telling me
00:25
<Hixie>
christ is that confusing
00:25
<webben_>
Hixie: ah okay.
00:25
<Hixie>
caps lock - b is a useful command, thanks
00:26
<webben_>
yw
00:26
<Hixie>
i don't understand why caps lock - a isn't the same thing though
00:26
<Hixie>
why two commands?
00:27
<webben_>
Hixie: Judging from http://www.freedomscientific.com/fs_products/JAWSkeystrokes.asp INSERT+A just reads the address bar
00:27
<Hixie>
capslock+a is say all in laptop mode
00:28
<webben_>
Hixie: I suspect say all is specific to buffers
00:28
<Hixie>
seems so
00:28
<webben_>
whereas in dialogs JAWS would be in some sort of mode that more closely follows MSAA
00:28
<webben_>
(like forms mode in webpages)
00:30
<webben_>
If you have a window which also has a document loaded into a buffer, I guess there is a logic in there being two different commands.
00:32
<Hixie>
yeah
00:47
<Hixie>
ok, here's a test case for JAWS: http://www.hixie.ch/tests/adhoc/html/screen-readers/001.html
00:52
<webben_>
What's it reading wrong there?
00:53
<Hixie>
reads everything after "elephants" on the second line in the same voice as "elephants"
00:53
<webben_>
Hixie: This is with the Web Rent-A-Crowd scheme or something is it?
00:54
<Hixie>
yeah
00:57
<webben_>
Hixie: what happens if you turn on announcements too? does it think there's more than one link there?
00:59
Hixie
hunts for announcements
01:00
<webben_>
Hixie: sorry, that's not necessarily the technical term
01:01
<webben_>
I just mean where it will say "unvisited link" or whatever before a link
01:02
<Hixie>
not sure how to enable that, but it does say "visited link" once on that page
01:03
<webben_>
well at least it only thinks there's one link
01:03
<webben_>
are you sure you've got the voices different enough that you'd be able to tell the difference?
01:03
<Hixie>
oh yes
01:03
<Hixie>
one is male and the other female
01:03
<webben_>
ah okay
01:04
<Hixie>
i love the way it reads the first line, too:
01:06
<Hixie>
"In the following text, only one word left paren the word. Quote elephants quote. Right paren! should be said like a link."
01:06
<Hixie>
though that depends on the synthesiser
01:07
<Hixie>
SAPI5 in particular seems to suck.
01:36
<Hixie>
weird, it'll read certain sentences differently when it's reading the paragraph vs reading just the sentence
02:00
<Hixie>
second screen reader test case: http://www.hixie.ch/tests/evil/screen-readers/002.html
02:32
<Hixie>
i'm amused by the alt text discussion
02:32
<Hixie>
the arguments are all about whether we should allow alt="" to be omitted
02:32
<Hixie>
the spec already requires alt=""
02:41
<om_food>
it does in some cases allow alt to be omitted
02:41
<othermaciej>
"In such cases, the alt attribute may be omitted"
02:42
<othermaciej>
allowing it to be omitted in some cases means conformance checkers can't report failure whenever it is missing
02:42
<othermaciej>
which may have some small effect at the margins
02:42
<othermaciej>
though whether that effect is good or bad is unclear
02:43
<Hixie>
iirc every case where it says may be omitted is followed by a should-level requirement that it be not omitted
02:44
<othermaciej>
well yes, it's (almost) always strongly encouraged, but it is not 100% mandated
02:45
<othermaciej>
I'm thinking that is probably ok, though it does make JAWS give worse-than-usual results in many cases apparently
02:57
<Lachy>
Oh no! people are asking for RDFa in HTML :-( http://lists.w3.org/Archives/Public/public-html/2007Sep/0120.html
02:57
<Lachy>
(btw, how did that sneak into the charter?)
02:59
<Lachy>
oh good, it says "whether it is also allowed in the classic HTML serialization ... is for the HTML WG to determine."
02:59
<othermaciej>
Lachy: it snuck in there is an example
02:59
<othermaciej>
Lachy: it used to be there as a feature to add
02:59
<hober>
fortunately, the charter doesn't require support for RDFa, but simply encourages a mechanism to support extension
02:59
<othermaciej>
"The HTML WG is encouraged to provide a mechanism to permit independently developed vocabularies *such as* Internationalization Tag Set (ITS), Ruby, and RDFa to be mixed into HTML documents" (emphasis mine)
03:00
<hober>
yay "such as" :)
03:00
<karlUshi>
rdfa++
03:01
<Lachy>
karlUshi, I don't believe RDFa is good for anything, it's way too complex and overengineered
03:01
<Hixie>
surely not
03:01
<Hixie>
it uses namespaces, it must be good!
03:02
<karlUshi>
yes lachy, you put the key thing here "believe"
03:02
<Hixie>
it also uses RDF, and we all know that Web authors are rushing to author RDF and are only being stopped by lack of support in browsers
03:02
<karlUshi>
troll fest :) funny
03:03
karlUshi
is enjoying relax in a comfortable seat
03:03
<othermaciej>
RDF stands for RSS Data Format, right?
03:03
<Hixie>
hah
03:03
<karlUshi>
othermaciej: you could do better, Really Dat Fucked
03:04
othermaciej
is shocked at karlUshi's scandalous language
03:05
<karlUshi>
yes I know othermaciej but I'am applying your freedom of expression rule :) I'm just a robot
03:05
<Hixie>
i'm going to complain to DanC right now!
03:05
Hixie
ducks
03:05
<karlUshi>
you should
03:06
<Lachy>
oh, it also uses CURIEs. that's even better ;-)
03:07
<Hixie>
isn't that a measurement of radioactivity
03:07
<hober>
I'm trying to come up with something sufficiently righteously angry-sounding to say, but am laughing too hard to think of anything.
03:07
<karlUshi>
Lachy: to see through
03:31
<Lachy>
hey Hixie, how does your clipboard.cgi script work? Does it just save POSTed data to a text file and then send it back on GET requests?
03:33
<Lachy>
Hixie, can I get a copy of it to run on html5.lachy.id.au for my own tools?
07:07
<Hixie>
Lachy_: yes
07:41
<Hixie>
wow these videos are really interesting
07:43
<othermaciej>
what videos?
07:44
<Hixie>
http://lists.w3.org/Archives/Public/public-html/2007Sep/0103.html
07:45
<othermaciej>
ah cool
07:45
<othermaciej>
yeah, actual user testing
07:54
<Hixie>
man this is so interesting
07:55
<Hixie>
in the first 10 minutes of the table video it seems the biggest problems are with a bug jaws has about colspan/rowspan
07:55
<Hixie>
also interesting with the budget table is that it's clear that the headers being read out are completely useless
07:55
<Hixie>
it's not reading the column header
07:57
<hsivonen>
Hixie: also, JAWS has pretty significant trouble with thousand separators that are commonplace
07:59
<Hixie>
hsivonen: yeah, the number handling is appalling in that instance
07:59
<Hixie>
notice he goes to the horrific configuration UI
07:59
<Hixie>
and does far better than I (a sighted user) did when I went there!
08:00
<Hixie>
i think that revenue table (http://www.wisc.edu/about/facts/budget.php#budgetRev) is an example of a table that is badly marked up with headers=""
08:00
<Hixie>
"source" shouldn't be on the headers="" list for the data cells
08:00
<Hixie>
and that table uses abbr="" in a really confusing way, too
08:01
<Hixie>
"amount state revenue source. 151.8 column 2" is not a useful way of showing that data.
08:04
<hsivonen>
Hixie: what would you do better (apart from striking "source" from the speech)?
08:05
<Hixie>
"source" is a header for first column, it shouldn't be given as a header for the data cells
08:05
<Hixie>
headers aren't transitive
08:06
<Hixie>
arguably the "state revenue" row shouldn't be a header for anything but the first column either
08:06
<Hixie>
though scope="" can't represent that while still keeping the row table-wide
08:06
<Hixie>
which is interesting
08:07
<hsivonen>
with VoiceOver, I found that due to lack of easy documentation and due to the aural UI features being undiscoverable without docs, it is really hard for a person who isn't a routine user to check if something works in a sane way
08:07
<Hixie>
i think you'd want that first data cell read out as "255.1 millions of dollars, general program appropriations, state revenue"
08:08
<Hixie>
second cell as "-1.7 percent change from previous year, general program appropriations, state revenue"
08:09
<Hixie>
though how you decide that the top header comes first, then the left header, then the sub-top header, i dunno
08:09
<Hixie>
on the next table you'd want it the other way around
08:09
<Hixie>
second data cell on http://broads-authority.gov.uk/boating/navigating/tide-tables.html you'd want to read out as "1.69 meters, high water, wednesday 5th, september"
08:09
<hsivonen>
I've only watched the first two videos so far
08:10
<hsivonen>
and I think they demonstrate that "row" and "column" aren't the real scopes but "right" and "down"
08:14
<Hixie>
yeah that's basically what html5 does
08:15
<Hixie>
(though it uses the row/column names)
08:15
<Hixie>
the federal bank table example is interesting, it's a table that uses scope but that shows jaws7 not supporting it
08:15
<Hixie>
it also shows that jaws has some heuristics to read out the first column even with no markup for it
08:16
<Hixie>
woah
08:16
<Hixie>
the immediate reaction to the first table with summary="" information is "i'm wondering how necessary is that."
08:19
<hsivonen>
also, it shows how the insistence on cultural variance in date and number formats is an accessibility problem
08:21
<Hixie>
yeah
08:21
<Hixie>
go <time>
08:21
<Hixie>
not sure what to do about numbers
08:21
<Hixie>
<n> seems excessive and unlikely to be widely used
08:21
<Hixie>
i'd rather test the waters with <time> first
08:21
<hsivonen>
there's no way normal authors who are accustomed to type in Word would stop and mark up their numbers as numbers
08:22
<Hixie>
yeah
08:22
<Hixie>
same could be said of dates though
08:22
<hsivonen>
I suppose having a comma, period or nbsp every three digits could match a heuristic pattern
08:23
<Hixie>
or apostrophe
08:23
<hsivonen>
but dealing with non-nbsp regular spaces would be hard
08:23
<othermaciej>
you could use lang to infer the likely number format
08:24
<hsivonen>
everything would be so much easier if all nationalistic vive la difference could be put aside here
08:24
<othermaciej>
or just try to recognize a bunch of widely used number formats
08:24
<othermaciej>
but I guess some are ambiguous
08:24
<othermaciej>
1.001 - is that just over 1, or just over 1000?
08:25
<Hixie>
you're not going to get france or the uk to change to 1,001 or 1.001 respectively
08:25
<Hixie>
(or vice versa)
08:25
<hsivonen>
othermaciej: to some extent yes. but there are a lot of people who use English together with ISO dates, SI units and 24 hour clock for international communications
08:25
<hsivonen>
(in fact, it annoys me that legacy English date-time stuff comes with the English UI language on many systems)
08:27
<hsivonen>
https://bugs.maemo.org/show_bug.cgi?id=1968 for instance
08:27
<othermaciej>
there's also wide disagreement on the right way to do quotation marks but that seems less of a practical problem
08:28
<hsivonen>
systems would benefit from a U.S. English plus metric/ISO "locale"
08:35
hsivonen
sighs at optimism about parsing text/html as XML
08:35
<Whiskey_M>
'lo
08:35
<hsivonen>
back when Netscape decided not to sniff XMLness in text/html in Mozilla, O'Reilly was ill-formed
08:36
<Hixie>
wow the longdesc="" comments are really interesting too
08:36
<hsivonen>
good luck with others getting it right if the xml.com folks don't
08:36
<Hixie>
hsivonen: http://www.damowmow.com/playground/html-or-xml.html ! :-D
08:53
<Lachy>
wow, I didn't realise screen readers were sooo bad at reading out numbers!
08:54
<Lachy>
I would have expected it to read 278,000 as "two hundred and seventy eight thousand", not "two seven eight zero zero zero"
08:55
<Whiskey_M>
it changes dependent on the screen reader I believe (and version of the software), but in general they aren't great
08:56
<Lachy>
hmm. I guess. So is it a problem that screen readers should really solve, or do we need a number element? like <n>278,000</n>
08:57
<Lachy>
I wonder how they handle numbers written using with the use of ',' (separator) and '.' (decimal point) reversed
08:58
<hsivonen>
is it known how much http://www.nvda-project.org/ sucks/rocks compared to JAWS?
08:59
<Lachy>
cool, I didn't know there was an open source screen reader available. I wonder how mature it is
09:00
<Lachy>
well, I guess it's only version 0.5, so probably not very mature
09:02
<virtuelv>
Lachy: is <n>278,000</n> "two hundred and seventy eight, comma triple-zero" or "two hundred and seventy eight thousand"?
09:02
<Hixie>
lachy: as far as i can tell that's a jaws bug that the experienced jaws user hadn't seen before
09:04
<hsivonen>
interestingly, both nvda and orca are written in Python. I wonder if there's any cross-pollination
09:04
<Hixie>
http://www.cs.auckland.ac.nz/~pgut001/pubs/malware_biz.pdf is frightening
09:07
<Lachy>
virtuelv: it's 278 thousand
09:08
<Lachy>
there'd need to be a space after the comma (278, 000) if it was two hundred seventy eight, triple zero
09:11
<karlUshi>
virtuelv: it depends on your language
09:11
<karlUshi>
in French comma is the separator
09:11
<karlUshi>
for decimal
09:16
<Lachy>
Thanks Hixie, I've now got http://html5.lachy.id.au/clipboard running :-)
09:18
<hsivonen>
Lachy: what is it supposed to do?
09:19
<Lachy>
that's a clipboard, just like Hixie has for the Live DoM Viewer
09:19
<Lachy>
POST data to it to save or GET to retrieve
09:19
<Lachy>
I'm just adding the upload/download scripts to my other tools now
09:20
<hsivonen>
Lachy: the use case is not obvious to me
09:22
<Lachy>
it makes it easier to save data and then download it again in other browsers or on other computers
09:22
<Dashiva>
It's a pastebin, basically
09:24
<hsivonen>
Lachy: does it expect form POSTs or raw entity body POSTs?
09:26
<Hixie>
raw
09:30
<Hixie>
interesting, the power jaws user in question hadn't come across [D] links
09:30
<Lachy>
that's not too surprising. I rarely see them myself
09:37
<Hixie>
his comments about a long _description_ being useless to him are interesting
09:37
<Hixie>
(as opposed to a long discussion of the image)
09:37
<Hixie>
seems to me it would be better to just put the long discussion in the alt="", though.
09:38
<Hixie>
there's a bug in JAWS' handling of newlines
09:38
<Hixie>
in attributes
09:43
<virtuelv>
Lachy: karlUshi was making my point for me. My question was largely a rhetorical one
09:43
<Hixie>
seems the summary and header_id videos are the same
09:43
<Hixie>
which is odd
09:43
<Hixie>
oh well
09:43
<Hixie>
i really should be in bed 3 hours ago
09:43
<Hixie>
bbl
10:02
<Lachy>
I've now implemented the clipboard scripts in http://html5.lachy.id.au/output and http://html5.lachy.id.au/content-sniffing/ so now you can share data between them easily
10:08
<hsivonen>
does someone remember off-hand if HTTP-methods should be compared case-sensitively or insensitively?
10:08
<hsivonen>
(on the HTTP server when examining incoming requests)
10:11
<hsivonen>
case-sensitive it is
10:11
<Lachy>
hsivonen: they're case sensitive and, I believe, all existing methods (GET, POST, etc.) should be uppercase
10:23
<hsivonen>
Lachy: yeah, thanks
10:35
<hsivonen>
is there a best practice for transferring long non-ASCII strings in HTTP headers?
10:35
<hsivonen>
or should I just steer clear and do something else?
10:36
<Lachy>
hmm. you could use %-encoding, but that's really only for URIs
10:37
<Lachy>
what are you trying to do?
10:38
<hsivonen>
Lachy: I'm considering whether I should allow validator parameters as HTTP headers as opposed to query string params in the POST Web service API
10:38
<Lachy>
what HTTP headers are you trying to use for that?
10:39
<hsivonen>
Lachy: Content-Location for doc=
10:39
<hsivonen>
Lachy: the others would have to be X-Foo-Or-Bar
10:39
<hsivonen>
the potentially non-ASCII value would be schema IRIs
10:40
<Lachy>
you might be better off just using a query string
10:40
<hsivonen>
which I could force to be URIs
10:40
<hsivonen>
well, let's see if I can get away with query string only
10:45
<Lachy>
if a query string doesn't meet your needs, you could also try POSTing JSON or XML or something n the request body
10:46
<hsivonen>
Lachy: that would be much worse
10:46
<Lachy>
ok
11:15
<hsivonen>
aargh. Form-based form upload with the servlet API is ridiculous
11:15
<hsivonen>
supporting raw entity body POST was super-simple, though
11:20
<hsivonen>
do browsers guarantee that multipart form submissions contain the fields in the order of the original form?
11:20
<Lachy>
yes, I think so
11:24
<hsivonen>
well, on a more positive note, perhaps Java developers will resist the new Web Forms 2.0 submission format less than developers using other languages as even multipart/form-data support isn't part of the platform API
15:52
<Lachy>
in Table5.wmv of those usability tests, we were told that the table uses axis="", but he never actually showed how or why that is actually useful for anything
16:00
<zcorpan>
indeed
16:08
<Lachy>
I like what the bind man says at the end of Longdesc_IDC.wmv, where he talks about how longdesc can be made more mainstream by making it useful to more people
16:21
<Lachy>
the [D] link example didn't look any more difficult for the user than the longdesc attributes did, which is interesting given the claims that it's a problem because it's not explicitly associated with the image
16:34
<Lachy>
oh, I spoke too soon. in the BoxModel video, [D] links are explained to him
16:40
gsnedders
wonders how to get an email to the majority of major browser vendors regarding HTTP response parsng
16:53
<Lachy>
gsnedders: you could ask people you know from each browser vendor to put you in contact with the appropriate people who work on HTTP stuff
16:56
<gsnedders>
Lachy: I don't really know anyone who works on either IE or FF, though
16:57
<Lachy>
gsnedders: maybe try Chris Wilson for IE
16:58
<Lachy>
and ask Hixie about Mozilla, he may know
16:58
<gsnedders>
Lachy: yeah, that what's I was thinking I would try — last time I tried Chris I never heard back, though
17:00
<zcorpan>
https://bugzilla.mozilla.org/buglist.cgi?query_format=advanced&short_desc_type=allwordssubstr&short_desc=&component=Networking%3A+HTTP&long_desc_type=substring&long_desc=&bug_file_loc_type=allwordssubstr&bug_file_loc=&status_whiteboard_type=allwordssubstr&status_whiteboard=&keywords_type=allwords&keywords=&bug_status=ASSIGNED&emailassigned_to1=1&emailtype1=exact&email1=&emailassigned_to2=1&emailreporter2=1&emailqa_con
17:00
<zcorpan>
l2=&bugidtype=include&bug_id=&votes=&chfieldfrom=&chfieldto=Now&chfieldvalue=&cmdtype=doit&order=Reuse+same+sort+as+last+time&field0-0-0=noop&type0-0-0=noop&value0-0-0=
17:00
<zcorpan>
...might give a hint about who are working on http in moz
17:00
<gsnedders>
ah. true. I'd looked through public bug trackers for what I'm interested in, nothing about it, though
19:06
<Hixie>
gsnedders: feel free to mail whatwg⊙wo about it if you want to hit browser vendors in general
19:14
<gsnedders>
Hixie: thanks
20:05
<Philip`>
http://www.w3.org/QA/2007/09/svg-open-2007-day-2.html - "SVG as texture for 3D Canvas" - hmm, that's possibly interesting in terms of making 3D applications that don't need to download dozens of megabytes of bitmap textures
20:55
<Philip`>
http://canvex.lazyilluminati.com/misc/3d.html - hmm, Opera 9.5 beats 9.2 quite measurably, but it still seems no browser is adequate at vector/matrix calculations in JavaScript :-(
20:56
<Philip`>
(That code was originally in a toy language that got compiled to JVM bytecode, which was far faster - I think it did 30fps easily on a slower computer than what I'm using now)
21:50
<jgraham>
Hixie?
21:50
<Hixie>
hey
21:52
<jgraham>
You pinged. I don't know if I can use the private message thing on freenode without an account
21:52
<Hixie>
are you on another network?
21:52
<Hixie>
or jabber/aim/icq/msn?
21:52
<annevk>
just use irc.w3.org
21:52
jgraham
is somewhat communications poor
21:53
<virtuelv>
jgraham: you need to register on freenode
21:54
<jgraham>
Hixie: I think anne's suggestion should work
21:54
<othermaciej>
jgraham: you need to register, or the person you are talking to needs to set up their account to accept privmsgs from unregistered nicks
21:54
jgraham
goes to register on freenode
21:55
<virtuelv>
/msg NickServ help
21:55
<virtuelv>
^^ for jgraham
22:39
<gsnedders>
jgraham: are you stalking me adding me on Flickr? :P
22:40
<gsnedders>
jgraham: you're the first person to add me from WHATWG, and I don't think we have any common friends at all :P
22:40
<jgraham>
gsnedders: Well if you will put photos on the internet, you can't expect people not to look...
22:41
<gsnedders>
jgraham: but look for photos by me? :P Twitter or my blog, I assume>?
22:42
<gsnedders>
jgraham: most people I've had add me I have explicitly pointed to my Flickr page, or know them well
22:42
<jgraham>
Er, I think I originally searched for your blog for some reason which I have forgotten.
22:42
<gsnedders>
heh
22:43
<gsnedders>
I think the only thing on my blog that links to it is photo of my sister + husband at their wedding
22:43
<jgraham>
Yeah but you're pretty much unique on google :)
22:44
<gsnedders>
wait, you mean that everything within the top 50 results being me makes me unique? :)
22:45
<gsnedders>
It's something absurd like that which is all me
22:46
<jgraham>
That would fit my definition of "pretty much unique", yeah.
22:47
<gsnedders>
here we go, result 59 isn't me
22:47
jgraham
is cursed/blessed with a much more common name
22:47
<gsnedders>
I'm not a pilot.
22:48
<jgraham>
<green-wing>Except on friends reunited</green-wing>
22:49
<Lachy>
jgraham, in python, what APIs are available for accessing the query string and post data of a request? I've looked at cgi.FieldStorage(), but AFAICT, it's only desiged for accessing name=value pairs and doesn't make it easy to get data in other formats
22:49
<gsnedders>
59, and 82 are the only 2 results in the top 150 that aren't me
22:50
<jgraham>
Lachy: Have you tried httplib2? It's not in the standard-lib but I think it's what you want
22:50
<Lachy>
ok, I'll look it up
22:50
<jgraham>
http://code.google.com/p/httplib2/
22:51
<Lachy>
it seems odd that that stuff isn't build into the language. like in PHP, I can do $_SERVER['QUERY_STRING'], etc. to get the raw data
22:53
<Philip`>
os.environ['QUERY_STRING']?
22:53
<jgraham>
Lachy: httplib is built in but httplib2 is supposed to be better
22:54
<jgraham>
Where builtin = standard-lib
22:54
<gsnedders>
ah well, I'm going, g'nite
22:54
<jgraham>
goodnight
22:54
<gsnedders>
I'll try to stay out of sweet shops, though, jgraham :)
23:08
<jgraham>
Lachy: Did you find what you wanted?
23:08
<Lachy>
I think so, os.environ gives me the query string from GET requests, and I think stdin will give me post data
23:10
<Lachy>
hmm. stdin didn't work for me
23:13
jgraham
wonders if there is a sane way of setting up python on dreamhost
23:13
<Lachy>
oh, it works. but not when I try to read stdin after I import cgi, which already reads it
23:17
<jgraham>
http://wordsandpictures.dyndns.org/tables/table_inspector.html now with "smart colspan" option
23:18
<annevk>
audio on ubuntu works again after a reboot
23:18
<annevk>
bizarre
23:18
<jgraham>
Er, I meant to link to the html5.org copy there
23:18
<annevk>
just like windows :)
23:18
<jgraham>
http://james.html5.org/tables/table_inspector.html
23:24
<annevk>
"smart colspan" meaning I can omit scope= entirely on my table?
23:25
<annevk>
feature request: what you probably also want is a view where all algorithms are applied so you can easily see the differences for a single table; no real suggestions for the UI though
23:54
<jgraham>
annevk: "smart colspan" is what Ben Millard described on the list. It allows you to emit scope="" on your table entirely, yes.
23:54
<annevk>
awesome
23:55
<annevk>
I was thinking about that when I wrote the table
23:55
<jgraham>
I agree about the multiple algorithms thing; I think I can make it work technically but I'm not sure how to present it either
23:55
<annevk>
Now in theory the first column should maybe have scope=row for each row but I don't really like that. I'm not sure how to solve that issue
23:56
<annevk>
(Don't really like the typing.)
23:56
<annevk>
Making the table XML well-formed so I could enter it in my blog system was already a pain
23:56
<annevk>
I should probably have used html5lib or something...
23:57
<takkaria>
fwiw, mobile IE seems to render pages OK though is rather terrible at fitting things on-screen decently
23:58
<jgraham>
annevk: I think special casing rows or columns that are entirely heading cells to be like scope="column" or scope="row" respectively should work
23:59
<annevk>
the problem is that they're not headings
23:59
<annevk>
well, they're both (as HTML4 seems to allow)