00:26
<Lachy>
Hixie, yt?
00:29
<Lachy>
Hixie, in your live dom viewer script, within the update() function, what's the purpose of using a timeout for this: setTimeout(updateDOM, 100); instead of just calling updateDOM(); directly?
00:33
<Philip`>
Is that the one that prevents the thing updating at more than 10Hz, to avoid thrashing your CPU if you're typing fast?
00:34
<Lachy>
no, there's a timeout within the updateInput() function for that purpose
00:37
<Lachy>
this timeout gets called after the parsing has already been done, and updateDOM() just updates the various outputs, such as the innerHTML and DOM tree views
00:38
<Lachy>
so it seems there is little to gain from using a timeout here, at least as far as I can tell
02:29
<Lachy>
hsivonen, yt?
02:39
<Lachy>
hsivonen, nevermind
02:55
<Lachy>
I got my local copy of the HTML Tools beta working with both browser based and HTML5 based parsing using hsivonen's parser. Now I just need to implement the UI for switching between them
09:45
<Hixie>
Lachy: so that any timeouts in onload handlers in the subdocument get processed before showing the dom
10:01
hsivonen
fishes the draft TAG finding on versioning did more to highlight how CSS has it right
10:01
<hsivonen>
s/fishes/wishes/
10:02
<hsivonen>
weird typos
10:37
<Hixie>
hsivonen: unfortunately people on the tag like versioning
10:38
<Hixie>
hsivonen: so they are unlikely to pay too much attention to what are the actual success stories (css, dom) and failures (js, html, xml)
10:46
<hsivonen>
Hixie: I think I'll take the time to comment anyway
10:54
<hsivonen>
Comparing CSS and HTML, the big problem with distibuted extensibility in HTML is that instead of an element having a set of properties/attributes on equal footing, one property--the element name--is special
10:55
<hsivonen>
so an element can't be -webkit-canvas, -moz-canvas, -o-canvasw -ms-canvas and canvas at the same time
10:56
<hsivonen>
also,the canvasness is less optional than rounded corners
11:03
<hsivonen>
I wonder how thing would be different if the element name was just another boolean attribute
11:03
<hsivonen>
so that <p> was syntactic sugar for <p="">
11:08
<Lachy>
hsivonen, there's a bug in livedom.validator.nu, it doesn't fire onload events
11:08
<Lachy>
http://livedom.validator.nu/?%3C!DOCTYPE%20html%3E%0A%3Cscript%3E%0Afunction%20init()%20{%0A%20%20w(%22init%22)%3B%0A}%3C%2Fscript%3E%0A%3Cbody%20onload%3D%22init()%22%3E
11:09
<hsivonen>
Lachy: yeah, known bug
11:09
<Lachy>
ok
11:10
<hsivonen>
Lachy: I think putting time into full emulation like that wouldn't be justified by the reason why I made the parser work in GWT
11:11
<hsivonen>
Lachy: however, I want to expand it to cover other purposes, but I can't justify speding time on those things right now
12:30
<Lachy>
hsivonen, I may have an easy way to implement it without too much trouble
12:31
<Lachy>
I also have a way to make it dynamically set quirks and standards mode
12:35
<Lachy>
hmm, it may only work in Firefox. I'll investigate it more
14:19
<Lachy>
damn, I hate working with IE8. It's painfully slow, incredibly unstable and throwing up bogus security warnings! :-(
14:20
<Lachy>
and every time it crashes, it clears my clipboard! wtf?
14:24
<Lachy>
hmm, it's not actually clearing the clipboard, but after it crashes, it has trouble pasting until I recopy.
14:24
<Lachy>
maybe that has something to do with running it in VMWare ThinApp inside VMWare Fusion
14:40
<Lachy>
ok, I found and diagnosed a crash bug in IE8. Is anyone here able to file a bug? Here's a test case http://lachy.id.au/dev/2008/ie8b2-crash.html
14:41
<Philip`>
Can't you file a bug?
14:43
<Lachy>
no, I never got given the ability.
14:43
<Lachy>
but I need to fix the TC, I just realised the pass condition is wrong, and revealed another bug in Firefox
14:44
<Philip`>
I think I still have invitation codes, if you want one :-)
14:45
<Lachy>
sure
14:46
<Lachy>
this is annoying. Thanks to IE8 continually trying to reopen a crashed page, it gets into a continuous crash and relaunch cycle :-)
14:57
<hsivonen>
does bug filing access come with some bad NDA stuff?
15:03
<Philip`>
There's not a real one, like on paper with signatures or anything
15:03
<Philip`>
I don't remember whether there's a checkbox saying "I agree not to tell people about blah blah blah"
15:04
<Philip`>
(As far as I can tell, the aim is just to improve the signal-to-noise ratio, not to keep anything secret)
15:08
jgraham
resists the temptation to suggest whatwg⊙wo as the html5 non-accessibility list
15:08
hsivonen
already resisted
15:09
<takkaria>
does that mean you're done with resisting and are now going to suggest that? :)
15:10
<hsivonen>
takkaria: no, I moved onto some more email
15:10
<gsnedders>
jgraham: According to my textbook, \frac{d\e^{3x}}{dx} = \e^{3x} * 3 per the chain rule — how does the power not decrease?
15:10
gsnedders
doesn't get how you do it through the chain rule
15:10
<Dashiva>
jgraham: It's true, though
15:11
<Dashiva>
gsnedders: d(e^a) -> d(a) * e^a?
15:11
<gsnedders>
Dashiva: yeah
15:12
<gsnedders>
Dashiva: it says you do that per the chain rule, but that would involve decreasing the power by one
15:12
<Dashiva>
Why?
15:13
<jcranmer>
d [ f(g(x))]/dx = f'(g(x)) * g'(x).
15:13
<gsnedders>
Well, (x+y)^a would be d(x+y) * (x+y)^{a-1}
15:13
<gsnedders>
*a
15:13
<jcranmer>
since d[e^x]/dx = e^x
15:14
<Dashiva>
But you're not derivating with respect to e
15:14
<gsnedders>
True.
15:14
gsnedders
headdesks
15:14
<jcranmer>
d[e^(3x)]/dx = f'(g(x)) * g'(x) = f(g(x))*g'(x)
15:14
<Dashiva>
It's more related to 3^x than x^3
15:15
<Dashiva>
And ln e = 1, so the log part cancels out
15:15
<Dashiva>
(or something, been some years since I did this)
15:15
gsnedders
missed all the teaching on differentiating exponents and logarithms
15:16
<Dashiva>
That would explain things :)
15:16
<gsnedders>
:)
15:33
<hdh0>
http://packages.ubuntu.com/intrepid/abrowser ubuntu debrands firefox
15:36
<Philip`>
hdh0: Am I right in understanding that the branded version of Firefox is still available, and abrowser is just an independent package for silly people to choose?
15:38
<hdh0>
yes, firefox-3.0-branding versus abrowser-3.0-branding
15:46
<jgraham>
Dashiva: Some people can't handle the truth
15:49
<gsnedders>
jgraham: What truth?
15:50
<jgraham>
gsnedders: see about 40 minutes ago
15:50
<gsnedders>
jgraham: Wrong. There is no spoon.
15:56
<jgraham>
gsnedders: Incidentailly "proving" the chain rule turns out to be not so hard, at least to my satisfaction
16:06
<Philip`>
"I'm Feeling Lucky" is reasonable text for a UI button, but it really doesn't seem appropriate for an API call (http://code.google.com/appengine/docs/images/imageclass.html#Image_im_feeling_lucky)
16:07
Philip`
wonders why they didn't just call it auto_adjust or something
17:00
<gsnedders>
Hmmmm.
17:01
<gsnedders>
\pi d-\frac{4000}{d^2} = 0 — solve for d.
17:01
gsnedders
can't remember how to do that at all
17:10
<Philip`>
Multiply both sides by d^2?
17:12
gsnedders
doesn't think this question should be so complex, and guesses his whole approach for solving it is wrong
17:14
<Philip`>
I'm not sure if I'm misreading or misunderstanding, but it seems trivial :-p
17:16
<Dashiva>
Is d a variable, or derivation?
17:16
<gsnedders>
Dashiva: Variable
17:16
<Dashiva>
looks like a regular quadratic
17:17
<Philip`>
Doesn't look like that to me
17:17
<Philip`>
Just multiply by d^2, add 4000, divide by pi, cube root, and maybe simplify a bit
17:17
<gsnedders>
I ought to start my physics homework, regardless of all this damned maths
17:18
<Philip`>
(Am I missing something obvious?)
17:18
<gsnedders>
Philip`: yes
17:19
<gsnedders>
Philip`: You end up with \frac{d^2+4000}{\pi} = d^3 when you say to take a cube root
17:19
<Philip`>
Are you thinking that 0*d^2 = d^2?
17:20
<gsnedders>
Philip`: yes :)
17:20
<Dashiva>
Oh, I thought it was pi/d, my misreading
17:20
<Philip`>
Hint: 0 != 1
17:20
<gsnedders>
Philip`: NO WAI
17:36
<Philip`>
(Unless I'm mistaken, I guess 0!=1 is true in general for any abstract algebra with more than one element, given the usual definitions of 0 and 1, or something, so it's not just a fluke of integer arithmetic)
18:21
hsivonen
wonders if Gandi is going to keep their pricing after beta
18:22
<hsivonen>
they seem to have globally the best RAM per euro deal
18:23
<hsivonen>
Dreamhost PS is actually pretty bad in terms of RAM per euro despite the exchange rate of the dollar
18:32
<Philip`>
Dreamhost PS doesn't even seem to give you root access, so you can't have fun compiling your own kernels and so on
18:33
<hsivonen>
Philip`: I haven't had a need to compile the kernel on my server. In general, I don't like becoming an accidental sysadmin
18:34
<Philip`>
That sounds quite sensible :-)
18:34
<hsivonen>
I compared prices on Xen hosting, and this is what I learned
18:34
Philip`
finds it odd that Dreamhost advertises a $5.95/month rate, which only applies if you pay for ten years in advance, which is almost as long as the HTML5 timetable - who's going to know they're still going to want a web server that far in the future?
18:35
<hsivonen>
my current host (kotisivut.com in Finland) has competitive RAM per euro pricing compared to anything except Gandi and Amazon EC2
18:35
<hsivonen>
Amazon EC2 does not scale down
18:35
<hsivonen>
Gandi isn't quite ready yet
18:35
<hsivonen>
Bytemark hasn't been able to keep up with Xen-based competition
18:35
<hsivonen>
Dreamhost PS is surprisingly uncompetitive
18:36
<hsivonen>
companies in the Netherlands and in the UK have insane prices compared to Finland, France and the US
18:36
<hsivonen>
so my current host seems very good compared to anything but Gandi
18:36
<hsivonen>
and Gandi seems almost too good to be true
18:37
<hsivonen>
anyway, I sent email to Gandi requesting a beta invite
18:37
<Philip`>
Have you found anything about differences in reliability and support?
18:37
<hsivonen>
no
18:38
<hsivonen>
but my current host doesn't say anything about CPU shares
18:38
<hsivonen>
while Amazon and Gandi say what you get
18:39
<hsivonen>
oh, and Amazon has totally different disk persistence characteristics
18:39
Philip`
presumes CPU is harder to advertise since it can be very bursty, whereas RAM can't
18:39
<hsivonen>
I'm also surprised that googling for Xen hosting didn't turn up any Swedish or Estonian offerings
18:40
<hsivonen>
perhaps they don't do business in English or something
18:47
<hsivonen>
I want to see this area of business evolve into a model where someone own the datacenters and provides Xen paravirtualization and block storage, then someone else provides as open source on top of that App Engine -like automatic scaling so that the system is commodity open source and you can move to a different service provider
18:48
<hsivonen>
I think Amazon SimpleDB and App Engine are services that should be there, but using that high-level software shouldn't lock me in to a particular data center owner
18:50
<hsivonen>
I wonder if there are some savings in Amazon running SimpleDB themselves instead of a third party running CouchDB on top of Amazon EBS and selling that service
21:21
<hsivonen>
it's been over a month since the last standards suck episode
21:23
<Dashiva>
Maybe they stopped sucking
21:29
<Lachy>
I uploaded the latest copy of the HTML5 Tools http://html5.lachy.id.au/beta/ - this one has both HTML5 and browser parsing in it
21:30
<Lachy>
but there's a bug I cannot figure out with the switching between the 2 modes
21:30
<annevk>
http://search.twitter.com/search?q=%22html%205%22 is also worth tracking
21:43
<hsivonen>
lots of people tweeting about one article...
21:46
<Philip`>
Are they tweeting because of the article, or tweeting because of other people tweeting about the article?
21:48
<annevk>
hsivonen, yeah, the last few are indeed not that useful
21:48
<annevk>
and arguably before that not too much either, but still, it's rather easy to follow
22:56
<tusho>
Why is it an error to explicitly close self-closing tags?
22:57
<tusho>
<link></link>, <input></input> etc
22:57
<annevk>
those are void elements and have a special syntax
22:57
<annevk>
i.e., <link> or <link/>
22:57
<tusho>
annevk: yes, but I don't see why it should be an -error- to close them explicitly
22:57
<tusho>
just optional
22:58
<annevk>
because people would be confused and think they can put stuff between those two tags
22:58
<annevk>
and because <br></br> means <b><br> rather than <br>
22:58
<annevk>
means <br><br>, sorry
22:58
<tusho>
annevk: it's just that i programmatically generate html and it's a lot easier to treat all tags equally
22:59
<nessy>
is that really the case? I don't seem to remember that being the general xml rule
22:59
<annevk>
nessy, it's the HTML rule :)
23:00
<annevk>
tusho, keeping a set around and doing a simple if x in set is not an option?
23:00
<tusho>
annevk: well, it is, but still :\
23:00
<nessy>
http://www.w3.org/TR/REC-xml/#sec-starttags <- has the <br</br> example
23:01
<nessy>
I think in this case we should stick with the general xml rule - this is totally non-intuitive and confusing
23:01
<annevk>
nessy, tusho, http://software.hixie.ch/utilities/js/live-dom-viewer/?%3Cbr%3E%3C%2Fbr%3E
23:01
<nessy>
it's also not the way it works in HTML4
23:02
<annevk>
nessy, in HTML4 <link></link> is an error too
23:02
<nessy>
really?
23:02
<annevk>
yes
23:02
<annevk>
http://www.w3.org/TR/html4/struct/links.html#h-12.3 "Start tag: required, End tag: forbidden"
23:03
<annevk>
HTML4 does not even allow <link/>
23:06
<nessy>
glad browsers are not that anal :)
23:06
<nessy>
I guess then it's a tradition
23:09
<annevk>
Hixie, the methods are still on HTMLFormElement?