00:41
<gsnedders>
heh, nice, segfaults running it with Xvfb on some test
00:55
<gsnedders>
oh, no, crash happens anyway
01:11
<AryehGregor>
Theoretically, xvfb shouldn't be detectably different from Firefox's point of view.
01:11
<AryehGregor>
Otherwise it wouldn't be much good for testing in.
01:11
<gsnedders>
The biggest difference I've seen with Xvfb (in dealing with it on a much larger scale than this) is it has way more memory leaks.
01:12
<gsnedders>
This crash in Firefox happens when the default encoding is set to UTF-8. Heh. I guess that makes it low priority.
01:15
<AryehGregor>
I had a horrible memory leak in X a version or two ago. It was using like a gigabyte of RAM.
01:15
<AryehGregor>
I had to reboot. :'(
01:16
<gsnedders>
AryehGregor: One of the biggest problems I've been having dealing with Opera's regression tracking system is memory leaks in Xvfb. And we restart it ridicuously often.
01:16
<AryehGregor>
Interesting.
01:17
<AryehGregor>
Too bad C is so horrible that way.
02:11
<AryehGregor>
TabAtkins, your Worms clone lets you reset your shot if you click while it's still airborne. So you can just fire, watch where it goes, adjust before it hits, and repeat until you win on the first round.
02:12
<AryehGregor>
(I guess you don't have videos of your talks.)
02:15
<AryehGregor>
TabAtkins, also, you should use hyphenator.js on your website, since you're using text-align: justify. Otherwise lines get really stretched out near long URLs and such.
02:18
<AryehGregor>
. . . also, are you using 512-bit Whirlpool hashes again? Sheesh, these URLs look like they must be jokes.
02:18
<AryehGregor>
(And you really need a feed for your blog.)
02:19
<AryehGregor>
(Also, I really like the styling for the little date box permalink thing. Looks very neat.)
02:19
AryehGregor
goes to bed instead of gibbering random comments
05:02
<ai_scott>
can i get help with an issue with implementation of canvas in js here?
05:07
<ai_scott>
http://pastebin.com/ZZU6zvXz im having the issue on line 35 from 33 any ideas?
05:25
<aho>
eh... what are you trying to do there anyways?
05:25
<aho>
looks pretty nonsensical to me
05:26
<ai_scott>
aho thx write a image to a canvas
05:26
<aho>
eh... yea... well just load an image and draw it then *shrug*
05:27
<ai_scott>
i want to modify it and then write to file via a podt to a php page\
05:27
<aho>
right... draw it into a canvas... and then turn it into an image
05:27
<aho>
note however that the same origin stuff applies
05:28
<aho>
you cant read image data from images which came from a different domain
05:29
<ai_scott>
$img = '/path/jpj.jpg';
05:29
<aho>
(you can load it... and then draw it... but the canvas then also gets tagged as write only)
05:30
<aho>
$.get('<?php echo $img;?> <- what? :>
05:30
<ai_scott>
$img = '/path/jpj.jpg';
05:31
<aho>
so... this script is first processed by php?
05:32
<aho>
(you do use php to generate js?)
05:32
<ai_scott>
no
05:32
<ai_scott>
i just use the path generated by php
05:33
<aho>
yes... part of your js code is generated by php
05:33
<aho>
(that's bad for caching and whatever)
05:33
<ai_scott>
this is a one time image editor for the backend of a site
05:34
<aho>
yes... i mean you should pass that path to your script... and not inject it into the script
05:34
<aho>
well... alright... everything is on the same domain, right?
05:34
<ai_scott>
i want to change the brightness and contrast yup
05:35
<ai_scott>
once i get writen to the canvas i can use Pixastic to mod it and a post to a php page to save it
05:36
<aho>
ok... don't use that data url there
05:36
<aho>
just the path to the image
05:36
<aho>
and then you got your image
05:37
<aho>
(all that binary -> b64 stuff doesnt make any sense)
05:39
<aho>
http://kaioa.com/k/ct/image_combine/index.html
05:39
<aho>
http://kaioa.com/k/ct/image_combine/image_combine.js
05:39
<aho>
i'm using a silly loader there... once i got the image i do stuff with it
05:40
<aho>
it's pretty straightforward, really
05:42
<aho>
well, the example isn't as straightforward as it could be... i'm creating images out of other images before i draw them
05:43
<aho>
so... in renderScene... i could just call ctx.drawImage(img_rgb, ...) for example. and the unaltered image would appear on the canvas
05:44
<ai_scott>
ctx.drawImage('/localpath/image.jpg',0,0); ?
05:45
<ai_scott>
or something like this
05:45
<ai_scott>
http://pastebin.com/4QEY4YKk
05:45
<aho>
no... create an image first
05:46
<aho>
drawImage takes an image (or another canvas... or a video) as parameter
05:46
<ai_scott>
image element?
05:46
<aho>
<ai_scott> http://pastebin.com/4QEY4YKk <- ye, something like that :>
05:49
<ai_scott>
Uncaught TypeError: Cannot call method 'getContext' of null
05:51
<ai_scott>
got it
05:51
<ai_scott>
but it seems streched
05:59
<ai_scott>
aho thx for your help
06:12
<aho>
ctx.drawImage(im, 0, 0); <- draws the image at 0/0 without scaling or clipping (i.e. it shouldn't be stretched)
08:09
<annevk>
almost publication day
08:10
<annevk>
anything besides <b> that I should mention in http://dev.w3.org/html5/html4-differences/ ?
08:11
<annevk>
oh, maybe the killing of PUT/DELETE on <form method>
08:11
<annevk>
everything else seems like tweaking details
08:11
<annevk>
and with <b> I mean <s>
08:38
<asmodai>
Mmm
08:39
<asmodai>
why would 4.0b7pre say there's updates available at moz.org/products/firefox (which is 3.6.x)?
08:51
<hsivonen>
asmodai: where does it say that?
08:52
<hsivonen>
asmodai: anyway, my best guess is that it's a bug
08:52
<annevk>
do you use an en-dash for flip-side?
08:52
<hsivonen>
annevk: no
08:52
<annevk>
things like same-origin policy?
08:52
<hsivonen>
annevk: no
08:53
<hsivonen>
the only case where the Chicago Manual of Style supports lengthening a compound word hypen into an en dash is if the first part of the compound word has two parts and is capitalized
08:54
<hsivonen>
as in Civil War–era
08:54
<hsivonen>
which to me seems like a CMoS invention and not like common usage
08:55
<hsivonen>
so you *might* be able to cite CMoS to get away with Same-Origin–policy if you capitalize Same-Origin
08:56
<annevk>
interesting
08:57
<jgraham>
You might want better things to worry about if this is what is keeping you awake at night, however
08:57
<annevk>
it's early morning
08:57
<jgraham>
And?
08:57
<annevk>
also, why would I want to be concerned with real worries?
08:58
<hsivonen>
annevk: sorry. I remembered wrong. CMoS says you can use the en dash for a compound adjective in place of a hyphen if part of the compound is an "open compound (such as New York)" or itself hyphenated
08:58
<jgraham>
For all I know you were tossing and turning last night unable to sleep for thoughts of en and em dashes
08:59
<hsivonen>
except the latter rule is fuzzier, so you you aren't permitted to use and en dash for non-English-speaking or not-to-be-forgotten
08:59
<hsivonen>
annevk: the chapter and verse is 5.117 in the 14th ed.
09:00
<hsivonen>
anyway, the use of an en dash for compound adjectives smells like a CMoS oddity to me
09:14
<asmodai>
hsivonen: in about > help
09:14
<asmodai>
hsivonen: in that pop up window
09:35
<hsivonen>
asmodai: I don't see that in b8pre
09:36
<hsivonen>
asmodai: I thought there aren't nightly builds of b7pre anymore
09:36
<asmodai>
hsivonen: mmm, shouldn't b7 be able to move to b8?
09:37
<asmodai>
via the normal updates?
09:37
<hsivonen>
b7pre should update to b8pre, yeah
09:37
<hsivonen>
unless maybe if you are on Mac OS X 10.5
09:37
<hsivonen>
dunno if the 10.5 update problems have been solved already
09:38
<asmodai>
Nah, win7
09:39
<asmodai>
weird, it directs me to the normal URL yet my channel is set, as it should, to nightly.
09:40
<asmodai>
time to fetch it manually I guess
09:43
<hsivonen>
asmodai: I tried on Windows 7: b7pre from September 20 or so autoupdated to b8pre
09:43
<asmodai>
hsivonen: peculiar
09:43
<asmodai>
hsivonen: 32 or 64 bit?
09:43
<asmodai>
I just installed via the installer and it offered the upgrade option. Now with b8 it says it's uptodate at least.
10:06
<asmodai>
mmm ie9b @ acid3 95/100 - html5 96+5, ff 4b8 @ 97/100 - html5 217+9, webkit nightly @ acid3 100/100 - html5 232+9
10:07
<hsivonen>
I'm guessing "html5" means html5test.com
10:07
<asmodai>
yep
10:07
<asmodai>
colleague on Mac and me test browsers on a semi-frequent basis since he is responsible for the uni's website stuff
10:08
<asmodai>
And right now some of our JS stuff in IE9 seems to go screwy :(
10:08
<miketaylr>
like what?
10:09
<asmodai>
miketaylr: looks like all the text in this "ticker" is being displayed all at the same time
10:09
<asmodai>
ok, and now reloading it shows it as it should.
10:09
<asmodai>
Mmm, lets chalk it up as a temp glitch
10:09
<hsivonen>
the bonus points on html5test.com are a bit questionable
10:10
<hsivonen>
bonus points for encumbered codecs, bonus points for SQL database
10:10
<jgraham>
s/a bit/very/
10:10
<hsivonen>
and OTOH treating SVG and MathML as bonus features rather than proper features
10:10
<asmodai>
that sucks
10:10
<asmodai>
I want my mathml support
10:11
<jgraham>
If it is going to treat SVG and MathML it whould have real tests though, not just superficial ones
10:11
<annevk>
we just released a build with Web Socket support: http://my.opera.com/desktopteam/blog/2010/10/11/websockets
10:11
<annevk>
also has our new EventSource implementation
10:11
<jgraham>
(insofar as any of the tests are not just superficial)
10:11
<zcorpan>
you should complain here http://github.com/NielsLeenheer/html5test
10:12
<jgraham>
annevk: Yay! Go us! ;)
10:12
<annevk>
#toocoolfortwitter
10:12
<annevk>
:)
10:12
<miketaylr>
\o/
10:14
<zcorpan>
yay
10:18
<asmodai>
ah right, need to update my opera too
10:18
<asmodai>
sorry anne!
10:18
asmodai
goes to add
10:28
<zcorpan>
just tested that http://html5demos.com/web-socket works with opera :)
10:28
<miketaylr>
now i can go back and enjoy some of the node.ko demos
10:38
<miketaylr>
lists.w3.org seems down
10:56
<asmodai>
that opera 10.70 is acid3 @ 99/100 - html5 179+9
10:58
<sideshow>
miketaylr: lists.w3.org seems to be working for me
11:00
<miketaylr>
ahh you're right, sideshow. back up.
11:15
<hsivonen>
so many blog posts I'd want to write but don't get around to
11:16
<annevk>
yeah me too
11:16
<annevk>
i'm trying to catch up today
11:16
<hsivonen>
today I'd want to write a blog post explaining why there aren't WYSIWYG, *direct manipulation*, non-sucking editors for CSS layouts
11:17
<hsivonen>
tl;dr version: the existence of reftests proves that the program can't know what CSS you want if you draw a constellation of boxes
11:18
<hsivonen>
maybe I should do my blog posts as two tweets each
11:22
<jgraham>
hsivonen: That isn't enough for a proof, because you could constrain the layout primitives used by the generator for uniqueness
11:23
<jgraham>
And uniqueness is not really necessary except in a weak sense
11:23
<jgraham>
(an action in the GUI should lead to deterministic results)
11:23
<hsivonen>
jgraham: more proof: MS Word
11:24
<jgraham>
I don't think word is a proof of anything in particular
11:24
gsnedders
screams and runs
11:24
<hsivonen>
jgraham: if you look at what layout parameters Word supports, it's surprisingly similar to CSS properties sans cascade and nesting
11:24
<jgraham>
although it might be evidence that it is a hard problem
11:25
<hsivonen>
jgraham: so you can get pretty good and complex layouts by spending a day typing numbers into boxes
11:25
<annevk>
it's hard because a good editor would infer meaning from style; but if the editor can do that, so can the browser...
11:25
<jgraham>
annevk: Meaning seems to have little to do with the question at hand
11:25
<hsivonen>
jgraham: but when you touch the stuff with the mouse and Word tries to guess what you meant, everything breaks and you become extremely frustrated
11:26
<jgraham>
annevk: That's the question of why editors typically generate bad markup
11:26
<hsivonen>
jgraham: if I draw a box, how's the editor to know what units or media queries or positioning/oating I have in mind?
11:27
<hsivonen>
*floating
11:27
<jgraham>
hsivonen: It has to constrain things somehow
11:27
<jgraham>
Or pick reasonable defaults and let you override them
11:28
<jgraham>
In any case the fact that Word, with a simpler model, does badly hardly constitutes proof it is *impossible* only evidence that it is hard
11:28
<annevk>
Brendan's take on versioning is interesting
11:29
<espadrine>
Where is it?
11:30
<annevk>
https://bugzilla.mozilla.org/show_bug.cgi?id=602028#c11
11:30
<annevk>
oops
11:30
<annevk>
https://bugzilla.mozilla.org/show_bug.cgi?id=602028#c17
11:33
<annevk>
I'm not sure I entirely agree though... while it certainly moves somewhat, the old CSS table layout stuff is still around
11:34
<annevk>
(quirks mode)
11:34
<jgraham>
It moves very very slowly
11:34
<annevk>
and versioning in JavaScript... well, I guess we will see what happens, but currently there is no clear that is going to work out of band
11:35
<hsivonen>
If had a chance to go back and redesign the Almost Standards vs. Standards CSS table thing, I'd make Almost Standards the default and make the Standards behavior an opt-in controlled by a CSS property rather than doctype
11:35
<hsivonen>
*If I
11:35
<annevk>
yup
11:36
<annevk>
at some point we'll get that property I think
11:36
<annevk>
when we define limited-quirks mode and quirks mode
11:59
<jgraham>
annevk: I hear that Safari 5.02 supports -76
12:04
<annevk>
oh really?
12:05
<jgraham>
Haven't tested or anything
12:05
<jgraham>
micheil claimed this
12:05
<gsnedders>
annevk: As far as I can tell, Harmony is going to have to be versioned too
12:06
<annevk>
http://html5demos.com/web-socket supports both versions I guess?
12:07
<annevk>
gsnedders, yeah he was talking about post-ES5
12:07
<annevk>
we'll see what happens
12:07
<gsnedders>
Nobody will use it.
12:07
<gsnedders>
This is my predication.
12:07
<hsivonen>
is there a non-racy way to detect if a meta Refresh navigation is under way?
12:07
<jgraham>
The ES people seem to have versioning mania
12:09
<annevk>
if it comes with speed benefit the large sites will use it
12:09
<annevk>
and libraries maybe
12:09
<annevk>
with their on the fly loading magic
12:09
<annevk>
but then the decision what to use will still be based on browser sniffing rather than versioning
12:34
<annevk>
guess I'll update html5-diff now
12:35
<annevk>
if anyone finds anything conclusive on Safari 5.02 and -76 of WebSocket that'd be great btw
12:35
<annevk>
I'd like to update my post if Safari also implements it
13:43
<zcorpan>
annevk: so safari 5.0 uses -75 but safari 5.02 uses -76
13:44
<zcorpan>
except it doesn't support the closing handshake, i think
14:43
<jgraham>
hsivonen: BTW I am warming to the proposal of converging on the WebKit script scheduling behaviour
14:43
<jgraham>
I worry that the "ordered" attribute is adding needless complexity however
14:45
<jgraham>
Do we have any data about how widely deployed require.js and LabJS are? Maybe if they update now and advertise that the new version is needed for Firefox 4 compat then everything will work out fine
14:50
<miketaylr>
That's an interesting question--Kyle, the author of LabJS mentioned somewhere in his blog that twitter.com and zappos.com use it.
14:51
<miketaylr>
And there's rumors that require will end up as part of Dojo 2.0...but I don't have any real evidence for that
15:53
<micheil>
jgraham: that would be correct. http://github.com/miksago/node-websocket-server/wiki/browser-compatibility
15:54
<micheil>
rather, url is: http://github.com/miksago/node-websocket-server/wiki/Browser-Support
15:58
<micheil>
zcorpan: do you know a rough release date for opera 10.70?
15:58
<micheil>
(that way I can set myself a reminder to test websocket support in it.)
16:00
<zcorpan>
micheil: i don't
16:01
<micheil>
fair enough.
16:04
<jgraham>
Pop quiz: I document.write an external script element into an iframe. Does the external script load before the document.write call returns?
16:04
<jgraham>
per spec
16:08
<annevk>
thanks micheil
16:08
<annevk>
for the Safari update
16:08
<micheil>
no worries
16:09
annevk
updated http://annevankesteren.nl/2010/10/websocket-protocol accordingly
16:09
<micheil>
annevk: feel free to update that github page.
16:10
<annevk>
looks correct to me
16:11
<micheil>
I'm just reformatting it a bit..
17:25
<TabAtkins>
AryehGregor: Yeah, I know about the reset-shot thing. I'm currently just changing players on impact. I should be (a) turning off shots once you've made one, and (b) triggering a turn end if your shot goes off the side of the screen as well.
17:26
<TabAtkins>
AryehGregor: Yeah, no videos. Our tech was a little low at this event.
17:26
<TabAtkins>
AryehGregor: Never heard of hyphenator.js, will check it out.
17:26
<TabAtkins>
AryehGregor: I added hAtom last night, and will put together an actual Atom feed soon.
17:26
<TabAtkins>
AryehGregor: Also I like Whirlpool hashes because they're silly.
17:27
<TabAtkins>
AryehGregor: And thanks, I like the styling too!
17:27
AryehGregor
verifies that TabAtkins has addressed all of his feedback, check-plus
17:28
<TabAtkins>
^_^
17:29
<jgraham>
TabAtkins: Your worms game isn't yet multiplayer using websockets
17:29
jgraham
wonders if all feedback written here gets addressed :)
17:30
<TabAtkins>
hsivonen: In addition to the non-uniqueness of layouts, there's the simple fact that CSS *does not have* appropriate layout primitives yet. Template Layout will make WYSIWYG layouts fairly easy.
17:30
<TabAtkins>
jgraham: Nuts to that.
17:30
<jgraham>
TabAtkins: You're thinking of squirrels, not worms
19:31
<TabAtkins>
Okay, so, AryehGregor, I have a problem. I want to ensure that documents I've written can't be bruteforce discovered by some random attacker. But I want a shorter url, too. Discuss. (Manually providing shorturls for my blogposts doesnt' scale well, especially since I accidentally wiped out my shorturl resolver last week.)
19:32
<annevk>
set the URL for public documents
19:33
<TabAtkins>
That makes too much sense, annevk.
19:33
<AryehGregor>
TabAtkins, brute-force discovery is unlikely unless either your names are extremely short or they follow a consistent pattern or someone is really determined.
19:38
<variable>
TabAtkins, rate limit access
19:38
<variable>
and IP block
19:44
<Philip`>
Password-protect non-public documents?
20:25
<TabAtkins>
Solution I decided on over lunch: Keep the ridiculously long uuids, but also have an algorithmic shorturl for public documents.
22:49
<Hixie>
hsivonen: yt?
22:52
<AryehGregor>
TabAtkins, http://validator.nu/?doc=http://www.xanthir.com/blog/
22:53
<TabAtkins>
Noted.
22:54
<Hixie>
i wish we could make -- valid
22:54
<Hixie>
do we really care that it's not mappable to XML?
22:54
<Hixie>
and that it causes problems in older Geckos?
22:54
<TabAtkins>
NO.
22:54
<TabAtkins>
That was more forceful than intended.
22:54
<TabAtkins>
No.
22:54
<Hixie>
heh
22:55
<TabAtkins>
So I've fixed the error I care about. The validator can continue to complain about the line of ------ I have in a comment.
22:56
<Hixie>
anyone understand what http://www.w3.org/Bugs/Public/show_bug.cgi?id=10841 is about?
22:56
<TabAtkins>
I think he wants the ability for captions to adjust the playback rate.
22:57
<Hixie>
i don't understand what that means
22:57
<Hixie>
how could captions do that?
22:58
<TabAtkins>
Through some additional switch embedded in-band, I presume.
22:58
<TabAtkins>
The intent is so that with, frex, descriptions, you can have extra time to read the description while it's running.
22:58
<TabAtkins>
But only when you have the description track turned on.
22:58
<KaOSoFt>
Hello. In HTML5, can I do something like <br /> without breaking validation?
22:59
<TabAtkins>
Yes, using a trailing slash on void elements is just fine. It just doesn't mean anything.
22:59
<Hixie>
TabAtkins: hm, interesting. thanks.
22:59
<TabAtkins>
Doesn't mean anything special, that is. It's the same as <br>.
22:59
<Hixie>
i wonder if he means that this should just be allowed in principle, or that it should be part of a specific caption format, or part of <track>, or something else
22:59
<cardona507>
sans the cruft :D
23:00
<Hixie>
oh well, i asked him to elaborate
23:00
<TabAtkins>
Yeah, that's fine. If I didn't already know what the bug was about I wouldn't be able to parse what he said.
23:05
<hober>
TabAtkins: http://microformatique.com/optimus/?format=validate&uri=http://www.xanthir.com/blog/
23:06
<hober>
also http://lukearno.com/projects/hatom2atom/?url=http://www.xanthir.com/blog/
23:07
<TabAtkins>
hober: That validator is broken. For one, it's telling me that things aren't specified multiple times. For two, it's telling me that entry-title, updated, and bookmark aren't specified, then listing the entry-title, updated, and bookmark. For 3, author is specified in the feed, outside of the entry, as allowed by the hAtom format.
23:08
<TabAtkins>
For the second link, are you getting anything? I get a Python error.
23:08
<hober>
boo. :( have you sent a bug report to the validator peeps?
23:08
<TabAtkins>
hober: Not yet. I'm just now critiquing it. ^_^
23:09
<hober>
re: hAtom2Atom, yeah, I'm seeing the same thing, which is why I ran your page through the validator.
23:09
hober
is impatient to subscribe to your blog :)
23:09
<paul_irish>
for real.
23:09
<TabAtkins>
Shortlinks first. Then a real atom feed for you people. Is there any standard location/filename for an atom feed?
23:10
<Hixie>
anyone got a good idea of a new example for http://www.w3.org/Bugs/Public/show_bug.cgi?id=10743 ?
23:10
<Hixie>
(that is, an example of a conformance requirement on authors that is present only to help authors avoid a brittle part of the language)
23:11
<hober>
TabAtkins: no, but you might as well mint the cleanest url possible (mine is /feed)
23:11
<Hixie>
i.e. another example for "Errors involving fragile syntax constructs" in http://www.whatwg.org/specs/web-apps/current-work/multipage/introduction.html#syntax-errors
23:11
<TabAtkins>
hober: Makes sense. I'll probably do /blog/atom, I guess.
23:11
<TabAtkins>
Hixie: Anything involving <p> autoclosing?
23:12
<hober>
unquoted href with Sufficiently Weird URL in it?
23:12
<Hixie>
i don't know that we have any conformance requirements to avoid that (the only thing that's really fragile there is <p> bla bla \n <ins> <p> bla bla , and nothing is especially done to help authors avoid that issue
23:13
<Hixie>
)
23:13
<Hixie>
hober: hmm, href=?a=b maybe
23:13
<Hixie>
hober: though that's really illegal for another reason (likely typo, not being fragile)
23:14
<hober>
something involving the AAA?
23:14
<TabAtkins>
Actually, that's probably a good example, since it's one of the errors the validator flagged on my site.
23:14
<TabAtkins>
(It wasn't a typo in my case, of course.)
23:14
<gsnedders>
For those who have me on Facebook and feel helpful: http://www.facebook.com/gsnedders/posts/448566623395
23:15
<TabAtkins>
gsnedders: What do you need help with?
23:15
<TabAtkins>
gsnedders: Seems like everything's taken care of.
23:17
<gsnedders>
TabAtkins: The specific post? Or does that not show?
23:17
<TabAtkins>
gsnedders: I see the post. You gave an answer, and the dude seems to have understood it.
23:18
<TabAtkins>
AryehGregor: Permalinks are now short. Happy?
23:18
<gsnedders>
TabAtkins: I'm too tired to try and understand people :P
23:20
<TabAtkins>
AryehGregor: (Using tantek's system of a 3-digit number encoding days since the epoch in a specially-crafted sexagecimal designed for maximum readability, followed by a sexagecimal digit encoding the index of the post on that day.)
23:23
<Dashiva>
I hope Mark writes the HTML4 deployment post
23:29
<TabAtkins>
This seems somehow wrong: <link>rel="self" href="/feed"</link>
23:29
<TabAtkins>
It's from a tutorial on the Atom format.
23:29
<TabAtkins>
Surely that should be <link rel="self" href="/feed" />?
23:59
<ai_scott>
TabAtkins surely