00:00
<Hixie>
o_O
00:01
<gavinc>
I think I have to remove some variables (using JQuery) and make this test case just use XHR, but seriously WTF
00:45
<Hixie>
MikeSmith: component watching used to be done years ago by having a component QA and then having people watching that component QA
00:45
<Hixie>
MikeSmith: that requires account watching to be enabled, though
07:57
<zcorpan>
TabAtkins: excellent. looks correct now afaict. thanks. i'll remove the quirks from quirks mode spec
10:08
<jgraham>
Huh, why am I having so much difficulty making chrome prompt me to stay on a page by using a beforeunload event?
10:16
<zcorpan>
TabAtkins: the asyncness of CAS means that there's a race condition -- elements might be parsed before the CAS is loaded, and do their thing (like, load something) before the CAS kicks in
10:17
<zcorpan>
i think we have enough trouble with race conditions as it is
10:39
<odinho>
jgraham: Isn't that stuff disabled for those? Was some talk about it at least.
10:40
<zcorpan>
prompting to stay is the reason beforeunload exists
10:41
<odinho>
zcorpan: Maybe it was unload if that is an event.
10:41
<zcorpan>
yeah
10:46
<jgraham>
I eventually fixed up the problem
10:46
<jgraham>
Firefox doesn't seem to prompt with history.go
10:46
<jgraham>
Chrome didn't like it when I used a function from a different document as the event handler
10:46
<jgraham>
(I don't think)
10:46
<jgraham>
(which seems like a bug)
10:47
<jgraham>
Well one bug each
11:00
<jgraham>
Oh, I lied gecko didn't have that bug
11:00
<jgraham>
So now I wonder if I was doing something strange
12:09
<ruby_on_tails>
hello
12:09
<ruby_on_tails>
if i draw an arc in canvas it starts from the rightmost, how can i make it start from the top
14:23
<crankharder>
I'm having a problem with the application cache in chrome. My app cache looks like this: https://gist.github.com/3426029 Once this has been loaded, the fallback section takes over for all non-200 responses, which means that in all error conditions (4/500), my app falls back to /offline, which is not expected behavior. Should this not be happening, or is there some way to avoid it?
14:25
<crankharder>
As "/ /offline" is a typical example in the docs, I can't imagine that it is recommended to circumvent all non-200 requests while the user is online.
15:12
<jgraham>
OK, now I know what tool I need for the spec
15:12
<jgraham>
I want something that marks up bugs filed on different bits of the spec
15:12
<jgraham>
So when someone submits a bug on a specific section everyone can see it
15:13
<zcorpan>
yes, bug annotations would be great
15:14
<jgraham>
I need this so much, jsut to track bugs that *I* filed I think I will have to implement it
15:15
<zcorpan>
\o/
15:28
<Ms2ger>
\o/
15:37
<jgraham>
Hmm, so I guess it would be useful to remove bugs once they are closed. Which means that having bzAPI on W3C bugzilla would be rather helpful
15:37
<jgraham>
MikeSmith: ^
15:54
<jgraham>
Hixie: Can I get the bug filing / status updating CGI scripts?
16:24
<MikeSmith>
jgraham: I wasn't aware that bugs could be removed
16:24
<Ms2ger>
MikeSmith, closed
16:25
<Ms2ger>
Then they shouldn't be listed in the in-spec annotations
16:25
<MikeSmith>
ah
16:25
MikeSmith
tries to recall previous discussions about bzAPI
16:26
<MikeSmith>
is it not already enabled?
16:26
<jgraham>
MikeSmith: Myabe?
16:27
<jgraham>
I didn't actually try, I just assumed it wasn't
16:27
<MikeSmith>
ah I think it's not
16:27
<MikeSmith>
because v1 was released in 2011
16:27
<MikeSmith>
and we are running 3.6 from 2010
16:27
<MikeSmith>
and it's an add-on I guess
16:29
<jgraham>
Yeah it's an addon. but it seems to be 3.6.x compatible
16:29
<jgraham>
At least for some value of x
16:30
<MikeSmith>
ok
16:40
<Ms2ger>
MikeSmith, I am told it does indeed support 3.6
16:40
<MikeSmith>
OK
17:39
<Hixie>
jgraham left :-(
17:39
<Hixie>
oh, no, it was just a netsplit!
17:39
<Hixie>
jgraham: ping
17:46
<jgraham>
Hixie: pong
17:46
<Hixie>
jgraham: i like your idea and wish to subscribe to your newsletter
17:47
<jgraham>
Heh
17:48
<Hixie>
jgraham: the way i've been doing similar things for my internal e-mail management is that i have a cronjob that just loads each bug one at a time and compares their current state, deleting e-mails relating to bugs not assigned to me or that are not closed
17:48
<Hixie>
jgraham: and so on
17:48
<Hixie>
jgraham: seems like we can do something similar here, just walking the bug list and reading the initial description of each one to get the section ID
17:48
<Hixie>
with some local caching to avoid hammering bugzilla it should be pretty simple to do
17:48
<Hixie>
(since you can just do a query to get the buglist, and then you only need to load the new bugs, bugs can't change section)
17:49
<jgraham>
Hixie: I was thinking that you would just hook up the bug submission thing to give you the initial list of bug ids
17:49
<jgraham>
But of course if you can get it from the actual bug data that would be better
17:49
<Hixie>
yeah you can totally just grab the comments and walk through them
17:49
<Hixie>
that's how i did the cloning, e.g.
17:50
<Hixie>
(bugzilla has an xml output mode)
17:50
<Hixie>
(so there's no scraping to do)
17:50
<jgraham>
Oh, that helps
17:51
<Hixie>
for both the query results and the bug data
17:51
<Hixie>
how were you thinking of exposing the data?
17:51
<Hixie>
simplest way for me is to just call a script on your domain and have you do whatever you want
17:52
<jgraham>
Exposing in what way? In the UI?
17:52
<Hixie>
yeah
17:52
<jgraham>
I was thinkking it should be part of the status annotations
17:52
<Hixie>
that works
17:53
<Hixie>
i can provide an API for you to hook into if you want, or we can do it all server-side and just have the status things come out of the database
17:53
<Hixie>
server side meaning you push teh data straight into the database rather than on each page load
17:53
<Hixie>
API meaning in the script
17:54
<Hixie>
i guess for you it's lighter load to do it via server-side comms
17:54
<Hixie>
and let me deal with the UI
17:54
<jgraham>
Yeah, that makes sense to me
17:54
<Hixie>
k
18:00
<jgraham>
OK, I can write the code to grab the data. What do I need to do to push it into the database?
18:01
<Hixie>
have your script output the data in the form "sectionid bugid bugid bugid\nsectionid bugid bugid\nsectionid bugid"
18:02
<Hixie>
i'll provide you a secret URL to POST to that will update teh DB
18:05
<jgraham>
So each time it will provide all the bug ids, and bugs that got resolved will be removed from the db on your end?
18:21
<Hixie>
jgraham: sounds good
19:06
<Hixie>
jgraham: should i store bug IDs, or bug URLs?
19:06
<Hixie>
jgraham: because another way i could do this would be just to change the current "demos" feature to a "bugs" feature
19:06
<Hixie>
jgraham: since we don't use the "demos" feature
19:06
<Hixie>
jgraham: this would also allow other people to add bugs and stuff
19:07
<Hixie>
jgraham: i guess it's easier to just have a list of IDs and not make it editable
19:13
<jgraham>
Hixie: I think have a list of IDs and if people want to edit it they can add a string like the one that the bug script produces as a comment in the bug
19:15
<Hixie>
makes sense
19:16
<Hixie>
hopefully nobody abuses it; if they do we can always have some specific people whitelisted who can use a "cancel" magic word :-)
19:33
<Hixie>
ok 33% complete on my end: the database has a bugs table and the cgi script passes the data to the client
19:33
<Hixie>
no way to add to it yet...
19:34
<jgraham>
I guess I will finish my end tomorrow; am feeling kind of sleepy now
19:34
<jgraham>
But it is coming together
19:36
<Hixie>
cool
19:37
<Hixie>
let me know if you need any help figuring out how to get xml out and stuff, or how the schema they use works, i still remember it all from doing the cloning
19:51
<Hixie>
is it me or is today a bad spam day
20:24
<hober>
Hixie: do you manually approve ml subscriptions?
22:04
<Hixie>
hober: no
22:05
<Hixie>
hober: why?
22:11
<hober>
Hixie: just wondering. a coworker's subscription took longer than he thought it would; probably just an email hiccup
22:19
<Hixie>
hober: weird
22:20
<Hixie>
hober: the only moderation action i take is with messages > 40kb, and those i mostly only allow if they're mine :-)
22:20
<hober>
*nod*
22:20
<Hixie>
(those that aren't usually are 40kb of quoted content with top-posting, which is against the mailing list etiquette)
22:21
<Hixie>
anyone remember what the bug is where i talked about appcache having a js-driven "server"?
22:21
<Hixie>
i can't find it
22:34
<Hixie>
Truncation operations cannot be performed if the session holds an active table lock
22:35
<Hixie>
wtf
22:36
<Hixie>
well, i guess jgraham better not ever call this reentrantly
23:11
<Hixie>
jgraham: ok, server-side is done, now working on client side.
23:22
<jacobolus>
re: http://www.w3.org/TR/compositing/#blendingnonseparable
23:23
<jacobolus>
is there some reason why Photoshop’s arbitrary and idiosyncratic compositing modes are being copied?
23:24
<Hixie>
jacobolus: because the editor is from adobe? :-)
23:24
<Hixie>
(client-side done as well, now i just have to debug it all...)
23:24
<jacobolus>
Hixie: I mean, I guess this is great if someone wants to use a browser to preview PSD documents
23:25
<jacobolus>
but these blend modes are based on what could be done quickly on early-1990s computer hardware
23:25
<Hixie>
you have to ask the editor
23:25
<zewt>
many of photoshop's blending modes are useful and if they're going to put them all under a patent disclaimer then cool, go for it
23:26
<Hixie>
(i don't think rik hangs out here)
23:26
<jacobolus>
doing a sum of .30 * R + .59 * G + .11 * B, for gamma-adjusted R, G, B, is based on the primaries used in the Adobe RGB space
23:27
<jacobolus>
but only actually properly yields luminance for linear RGB, not gamma adjusted
23:27
<Hixie>
it seems unsurprising to me that a spec from someone at adobe would be based on adobe's stuff...
23:27
<zewt>
(it's also the most useful thing if you're handed a PSD from an artist using random blend modes and told to recreate it dynamically)
23:28
<jacobolus>
the word “luminance” is something that was made up by Adobe, is given several incompatible definitions in photoshop, and has no external technical meaning
23:28
<jacobolus>
and their definition of “saturation” here is also not used the same way in any other context
23:28
<zewt>
(hint: opengl uses it all over the place)
23:28
<zewt>
(also hint: complaining in here isn't going to do anything :)
23:29
<jacobolus>
opengl is also based on early-1990s photoshop? bleh.
23:30
<Hixie>
if you just want to get it off your chest, please do go on, ranting here is always welcome. :-) but as zewt says, if you want to affect the spec, your time may be better spent e-mailing the spec's editor or the lists mentioned in the spec's status section
23:30
<Hixie>
(please don't take this as a suggestion to shut up though)
23:30
<jacobolus>
:)
23:31
<jacobolus>
I guess I don't really care what goes in the spec. not like there isn't plenty of other legacy garbage in the web tech stack
23:31
<cabanier>
hey
23:31
<othermaciej>
jacobolus: I don't really fully understand blend modes or why they are separate from compositing operators
23:31
<cabanier>
you don't like the spec?
23:32
<jacobolus>
othermaciej: those two terms are synonyms in most contexts :)
23:32
<jacobolus>
othermaciej: “blend mode” is photoshop’s word for it
23:32
<othermaciej>
jacobolus: well, not in this spec or in photoshop
23:32
<othermaciej>
they are sort of orthogonal
23:32
<jacobolus>
..?
23:32
<othermaciej>
(maybe I am wrong about photoshop)
23:32
<cabanier>
read the specification. It talks why they are different
23:33
<othermaciej>
you can have a blend mode of multiply and a compositing operator of dest-atop
23:33
<cabanier>
blending = how colors combine
23:33
<cabanier>
compositing = how you treat alpha
23:33
<Hixie>
oh hey, rik _does_ hang out here
23:33
<Hixie>
go figure!
23:33
<Hixie>
hey rik!
23:33
<cabanier>
:-)
23:33
<jacobolus>
othermaciej: fine. that's kind of irrelevant to my point, and that's a kind of idiosyncratic set of definitions, but it's no big deal
23:33
<zewt>
(also it's 2012, can we not put "1.0" at the end of spec names)
23:33
<Hixie>
jacobolus: i take it back, ranting here is productive after all :-)
23:33
<othermaciej>
a blend function takes src and dest color components as input, and gives a color (ignoring alpha) as output
23:34
<othermaciej>
so Cblended = B(Cs, Cd)
23:34
<othermaciej>
compositing functions also take into account alpha
23:34
<othermaciej>
so Ccomposited = C(Cs, As, Cd, Ad)
23:34
<othermaciej>
and if you have both, then you use the blend output as the source input to the compositing operator
23:35
<othermaciej>
which strikes me as somewhat weird
23:35
<jacobolus>
othermaciej: okay, let me rephrase my original comment then... is there some reason why Photoshop’s arbitrary and idiosyncratic compo^H^H^H^H^Hblend modes are being copied?
23:35
<cabanier>
yes
23:35
<cabanier>
they are widely implemented
23:36
<othermaciej>
jacobolus: I find the whole thing mysterious, but I am not enough of a graphics expert to know if there is a better approach to this sort of thing, or if this is the industry standard
23:36
<zewt>
jacobolus: another hint: everyone who works long in graphics (of any type, including ui) has an understanding of them
23:36
<cabanier>
PDF, Skia, coregraphics, win8 direct2d,
23:36
<cabanier>
Firefox
23:36
<cabanier>
etc
23:36
<jacobolus>
cabanier: "luminosity", "hue", "color" etc. modes are implemented all those places?
23:36
<cabanier>
they were also in the SVG spec (but were never implemented)
23:36
<jacobolus>
I find that hard to believe
23:36
<cabanier>
core graphic and win8 has them for sure
23:37
<jacobolus>
okay, apparently they're in PDF
23:37
<cabanier>
What is it that you don't like about them?
23:37
<zewt>
implemented color burn and several others in opengl a long time ago, heh https://svn.stepmania.com/svn/trunk/stepmania/Data/Shaders/GLSL/Color%20burn.frag
23:37
<jacobolus>
cabanier: they're arbitrary and produce bad results
23:37
<cabanier>
Designers already know about them
23:38
<cabanier>
standard teaching in design schools
23:38
<jacobolus>
they should have been fixed in photoshop as soon as the hardware could keep up, which was about 10 years ago
23:38
<cabanier>
jacobolus: do you have more info? Why do they produce bad results?
23:38
<jacobolus>
now unfortunately people are going to be stuck with technical choices made for the constraints of 20 years ago, into the foreseeable future
23:38
<Hixie>
like 80 character line length limits? :-D
23:38
<zewt>
Hixie: *rage*
23:38
<cabanier>
they were introduced in 98 or so
23:39
<Hixie>
zewt: but what if i want to save my code to punchcard!!!!
23:39
<cabanier>
please let me know if you have a better idea
23:39
<jacobolus>
cabanier: because .30R + .59G + .11B, when R, G, and B are gamma-adjusted sRGB, is an entirely nonsensical quantity
23:39
<zewt>
(if I see one more person writing python in 80-columns because they unquestioningly followed the drek that is PEP-8 i will launch them into space)
23:40
<gavinc>
ASCII being the most annoying, like no one thought maybe Iñtërnâtiônàlizætiøn would be important
23:40
<cabanier>
so, you don't like the non-separable ones?
23:40
<jacobolus>
they don't correspond to human vision, they don't correspond to any technical device
23:40
<jacobolus>
they're quantities made up by adobe for the constraints of early 1990s computers
23:40
<cabanier>
it's a reasonable formula.
23:41
<jacobolus>
no, it's not a reasonable formula
23:41
<cabanier>
As long as everyone implements it the same, the designer's choice will look the same
23:41
<jacobolus>
it's a formula that tells people (by the names used) that it's doing one thing, while in fact it does another
23:41
<cabanier>
there is no arguing about color math
23:41
<jacobolus>
that's great. but it sucks for all the designers involved, who end up with unintuitive dreck
23:41
<cabanier>
you think they are doing the math in their heads?
23:41
<zewt>
jacobolus: the point is that artists want something to look a certain way, and they use blend functions to achieve the look they want; that's their primary purpose
23:42
<cabanier>
they just used their tools/browsers until they like what they see
23:42
<jacobolus>
zewt: you got it. but my point is, what they want is not actually what they're getting
23:42
<zewt>
they're looking at what they're getting and going "yep. looks like what I want." which means they have what they want. that's really how it works :)
23:42
<jacobolus>
zewt: I mean, okay, in some sense that's true
23:43
<zewt>
it's the basic gap between artistic design and technical design
23:43
<cabanier>
people don't like that you muck with their colors. Adobe has made that mistake in the past.
23:43
<jacobolus>
it's a basic gap caused by shitty programming, because the dimensions involved don't correspond to human vision
23:43
<gavinc>
jacobolus: err, I don't think those are Adobe's those at least look like sRGB's numbers
23:44
<cabanier>
We've given up on most of the complex color math if you use default setting
23:44
<jacobolus>
gavinc: they're based on the primaries used by the Adobe RGB color space
23:44
<jacobolus>
gavinc: but don't make any sense for gamma-corrected components
23:44
<gavinc>
jacobolus: Are you sure? Those look like the primaries for sRGB
23:44
<jacobolus>
gavinc: yes, I'm sure
23:44
<gavinc>
jacobolus: okay :D
23:45
<jacobolus>
gavinc: the equivalents for the primaries used by sRGB would be .21 * R + .72 * G + .07 * B
23:45
<gavinc>
jacobolus: ah, yes. my mistake
23:46
<zewt>
(no, the gap between technical and artistic work is that one generally has a "right answer" and the other is entirely subjective; this applies fairly universally)
23:47
<jacobolus>
zewt: if I sat you down with image editing software that was properly based on human-relevant color dimensions, and then put you back on current software, you'd throw a fit, because there is a night and day difference
23:48
<cabanier>
jacobolus: please post on www-style if you have ideas on making the spec better.
23:48
<jacobolus>
cabanier: I think it's a lost cause
23:48
<Hixie>
were there people other than hsivonen who were against the way <template> parses into a separate doc?
23:48
<cabanier>
jacobolus: we've been there. everyone hated it
23:48
<cabanier>
jacobolus: the vast majority of people didn't understand it
23:49
<cabanier>
jacobolus: hardly anyone could print it
23:49
<jacobolus>
cabanier: basically, the quick decisions made by some guys at adobe one afternoon are destined to live on to eternity :)
23:49
<jacobolus>
what is "it"?
23:49
<jacobolus>
people didn't understand what?
23:50
<zewt>
jacobolus: maybe, maybe not--it doesn't matter because every artist I've ever worked with uses PS and expects output to be what they see
23:50
<cabanier>
jacobolus: device independent workflows. Everything icc based
23:50
<cabanier>
zewt: exactly!
23:50
<jacobolus>
cabanier: what does ICC profiles have to do with anything?
23:50
<zewt>
cabanier: btw. no vivid light? heh
23:50
<zewt>
(guess it's been a long time since I've been handed a PSD using that)
23:51
<jacobolus>
for anyone keeping score, I drew some nice diagrams and so forth at http://en.wikipedia.org/wiki/HSL_and_HSV which calls the dimensions involved in these blend modes “chroma” and “luma”
23:51
<jacobolus>
though again, those aren't ideal names
23:51
<cabanier>
zewt: I keep asking people if they want all the PS blend modes but I get no feedback
23:52
<zewt>
https://svn.stepmania.com/svn/trunk/stepmania/Data/Shaders/GLSL/Vivid%20light.frag don't recall what effect i implemented that for, though
23:53
<jacobolus>
cabanier: do you have any ins w/ people working on photoshop? can you tell them to add "exclusion" blend mode for files in CIELAB mode?
23:53
<jacobolus>
the combination of linear light mode + exclusion mode is one of the most powerful tools in photoshop (albeit entirely unused by anyone)
23:54
<gavinc>
possibly horribly wrong statement: I'm pretty sure .30R + .59G + .11B is a great deal older then anything at Adobe my memory (which has already proved faulty) is that it's from YUV used in PAL and NTSC but I may be totally wrong on that, my color math is bit a rusty and was mostly for print
23:56
<cabanier>
From the postscript (!) manual: the gray value for a given RGB value is computed according to the NTSC video standard. This standard determines how a color television signal is rendered on a black-and-white television set.
23:56
<gavinc>
Okay, NOT crazy
23:56
<cabanier>
you were right on! impressive!
23:56
<jacobolus>
gavinc: NTSC and Adobe RGB use similar primaries
23:56
<jacobolus>
not quite identical
23:57
<jacobolus>
gavinc: but that's beside the point
23:57
<jacobolus>
which is that images on the web are specified to be sRGB
23:58
<jacobolus>
and that adding linear combinations of gamma-encoded quantities yields nonsensical results
23:58
<gavinc>
okay, I think I mostly agree with that. But would need to put a great deal more color math back in my head before agreeing fully