| 01:36 | <annodomini> | Are there any groups working on standardizing any sort of web archive format (bundle of web pages and resources that can be distributed as a single file)? |
| 01:37 | <torchie> | so does <canvas width="100" height="100"> make a canvas that fills up the whole view or does that need percentages |
| 01:38 | <annodomini> | From what I can tell, Safari has its own web archive format, Microsoft has its own, KDE has it's own, and there's ePub for a specialized niche, but nothing general purpose. |
| 01:46 | <Philip`> | torchie: That is the size of the canvas bitmap in pixels |
| 01:46 | <torchie> | oh |
| 01:47 | <Philip`> | You'd have to use CSS if you want to stretch it to percentage sizes |
| 01:47 | <Philip`> | (but that won't make the bitmap have more pixels, so it won't look very good) |
| 01:47 | <Philip`> | (so you'd have to write some script to resize and redraw the canvas bitmap as necessary) |
| 01:47 | <torchie> | well what I wanted to do was make something that scaled depending on the size of the window |
| 01:47 | <torchie> | oh |
| 01:49 | <Philip`> | It can't be done entirely automatically, because the browser can't redraw the bitmap at a higher resolution, so you have to manually script it |
| 01:49 | <annodomini> | If you want something that scales with the window, you might want to look into SVG. Otherwise, you're going to have to listen for resize events, and set the width and heigh in pixels to the size of the window. |
| 01:54 | <torchie> | is <canvas> raster? can you render an svg graphic onto it? |
| 01:58 | <annodomini> | Canvas is a raster drawing surface with various raster and vector drawing operations for drawing to it. |
| 01:58 | <Philip`> | Some browsers let you put SVG in <img> and pass that to drawImage |
| 02:00 | <annodomini> | What I was referring to was simply using SVG, either inline in the HTML (currently only works in XHTML) or in an <object> or <img> tag. |
| 02:01 | <annodomini> | See http://emacsformacosx.com/ for a good example of scalable SVG being used to fit the whole window. |
| 02:01 | <othermaciej> | annodomini: that page is cool! |
| 02:01 | <othermaciej> | I mean, maybe I am just easily impressed but I dig the scaling |
| 02:02 | <annodomini> | Yeah, the scaling impressed me. |
| 02:03 | <torchie> | hmm |
| 02:03 | <torchie> | only some browsers, huh |
| 02:04 | <annodomini> | Canvas and SVG are not currently supported in IE. |
| 02:04 | <torchie> | doesn't osx already come with emacs |
| 02:04 | <annodomini> | It does, but just a command line version; this is a build of the cocoa version. |
| 02:05 | <torchie> | wouldn't that count as 'extras' and 'nonsense' |
| 02:05 | <annodomini> | I'm not sure I follow. |
| 02:05 | <torchie> | I thought emacs was primarily command line |
| 02:06 | <annodomini> | It's useful to be able to select text with the mouse, use more varied fonts, and get better keybinding by not running within Terminal.app |
| 02:07 | <annodomini> | Emacs has lots of frontends; a terminal based one, X based ones, Cocoa, Windows, etc. |
| 02:07 | <torchie> | i see |
| 02:08 | <torchie> | I'll check this out then; I've been using textwrangler lately |
| 02:08 | <annodomini> | Also, on the "only some browsers" point, there are projects for adding canvas & svg support to browsers that don't have them. |
| 02:09 | <annodomini> | There's excanvas http://excanvas.sourceforge.net/ which adds canvas support to IE by using VML, IE's proprietary vector language. |
| 02:09 | <annodomini> | And there's svgweb http://code.google.com/p/svgweb/ for adding SVG support to any browser that doesn't have it, by using Flash to render SVG. |
| 02:10 | <annodomini> | Both are going to have a few rough edges, but can be used for compatibility in a pinch. |
| 02:10 | <torchie> | I'm wondering how things will look for a canvas on larger and smaller screens |
| 03:25 | <TabAtkins> | annodomini: I'll be using excanvas for a project I'm working on. It performs great for what I need (generating graphs). |
| 09:02 | <aho> | why isn't there a clear() function in the canvas specs? clearRect(x,y,w,h) looks awfully longwinded... canvas.width = canvas.width is easier, but that kind of code is very weird, because everyone who didn't take a very close look at the specs will assume that this line doesn't do anything (but it clears the canvas) |
| 09:03 | <aho> | and why the f- aren't the push and pop functions called... y'know... push() and pop()? |
| 09:03 | <aho> | void save(); // push state on state stack |
| 09:03 | <aho> | void restore(); // pop state stack and restore state |
| 09:03 | <aho> | i mean... seriously... :> |
| 09:04 | <aho> | it *IS* a stack |
| 09:18 | <Hixie> | aho: ask whoever first invented the api at apple :-) |
| 09:21 | <aho> | i also think it would be nice if there would be some flag/hint for indicating that you're going to overdraw everything. the implementation won't have to use some buffer then (eg fbo in opengl) which would be a tad faster |
| 09:21 | <aho> | even more so if no fbos are available |
| 09:22 | <aho> | eg with the gles stuff on this embedded thingy here i don't even have fbos available :f |
| 09:23 | <aho> | so, the canvas specs are already set in stone? |
| 09:33 | <Hixie> | aho: what exists now is pretty stable, but i'm sure we'll add things over time |
| 09:33 | <Hixie> | i doubt we'll be changing what is there already though |
| 09:33 | <Hixie> | there are pages that depend on it already |
| 09:34 | <aho> | well, push and pop can be added.... save and restore can be kept (deprecated) |
| 09:35 | <Hixie> | we can never remove them, so just having two ways of doing the same thing just adds bloat for no good reason |
| 09:36 | <aho> | save/restore suggests a single state |
| 09:36 | <aho> | save/restore is different from other similar apis |
| 09:36 | <aho> | how does webgl call it? :> |
| 09:37 | <aho> | is the draft spec finally available (it was supposed to be released to the public in october) |
| 09:37 | <aho> | ? |
| 09:41 | <aho> | and well... those save/restore comments basically read like "yea this should have been called push... and this one over there is actually pop" :> |
| 09:54 | <Hixie> | aho: html5 has been released to the public since 2003 |
| 09:54 | <Hixie> | aho: http://whatwg.org/html5 |
| 09:55 | <aho> | webgl draft spec |
| 09:55 | <aho> | .) |
| 09:55 | <Hixie> | aah |
| 09:55 | <Hixie> | dunno about webgl |
| 09:55 | <Hixie> | i'm waiting myself |
| 09:56 | <aho> | heh |
| 09:59 | <aho> | i'll probably need a new graphics card to try it |
| 09:59 | <Philip`> | You'll probably need one that supports OpenGL 2.0 |
| 09:59 | <aho> | theoretically this one should be able to do it... but... well... it's ati. the ff nightly stuff only works in sw mode |
| 10:02 | <aho> | (fwiw it claims to support 2.1) :f |
| 10:08 | <aho> | ye, ezquake says gl_version 2.1.xxxx release |
| 10:08 | <aho> | should be good enough... in theory |
| 10:11 | <Philip`> | Hmm, maybe FF relies on other extensions that ATI doesn't implement? |
| 10:11 | <Philip`> | Seems like a bug if it doesn't work on a reasonably modern graphics card like that |
| 10:15 | <aho> | well, ati means bad drivers... they also stopped supporting it a few months after the release |
| 10:35 | <Dashiva> | I wonder what creates this overwhelming hatred for the design principles in some people |
| 10:38 | <aho> | ?:) |
| 10:57 | <othermaciej> | aho: in the context of graphics, it's pretty normal to speak of saving and restoring graphics contexts, rather than pushing and popping |
| 10:58 | <othermaciej> | (even though the saving and restoring follows a nested, stack-like model) |
| 10:59 | <ako> | push and pop are usually called push and pop, because things are pushed onto a stack or popped from one |
| 10:59 | <ako> | in opengl it's also push and pop |
| 11:00 | <ako> | so... webgl will most likely call these things also push and pop |
| 11:00 | <othermaciej> | in CoreCraphics, it's called CGContextSaveGState and CGContextRestoreGState |
| 11:00 | <ako> | to me it looks like they added the ability to save the transformation... and later on decided it might be nicer if it were a stack instead |
| 11:00 | <othermaciej> | <canvas> was originally modeled on the Mac OS X CoreGraphics API |
| 11:01 | <ako> | what's coregraphics? some apple thing? |
| 11:01 | <ako> | ye, well... that doesnt count then |
| 11:01 | <ako> | :) |
| 11:01 | <othermaciej> | it's the 2D graphics API of Mac OS X |
| 11:01 | <jgraham> | "let queue be a stack of elements" -- that's kind of confusing |
| 11:01 | <jgraham> | (Nothing to do with canvas I should note) |
| 11:02 | <othermaciej> | well, since <canvas> was originally invented by a team at Apple in the browser that ships with the Mac, it made sense at the time |
| 11:02 | <othermaciej> | personally I would have added a noun to the verb, since the naked verbs don't make it clear what is being saved and restored (or pushed and popped) |
| 11:02 | <othermaciej> | in that sense CG's SaveGState is more clear |
| 11:04 | <Dashiva> | jgraham: Java's Deque class is a better stack than the Stack class :) |
| 11:04 | <Dashiva> | Well, ArrayDeque to be specific |
| 11:04 | <zcorpan> | hello everyone |
| 11:04 | <jgraham> | zcorpan: hello |
| 11:05 | <othermaciej> | for what it's worth, CanvasRenderingContext3D doesn't seem to have eihter push/pop or save/restore |
| 11:05 | <othermaciej> | at least not the current copy in WebKit |
| 11:08 | <ako> | there is no push pop stuff in gles 1.x or 2.0 |
| 11:08 | <ako> | hum |
| 11:08 | <othermaciej> | hmm, I only just noticed that CanvasRenderingContext2D has readonly attribute HTMLCanvasElement canvas |
| 11:09 | <othermaciej> | amusingly this is retained in the breakout spec verbatim |
| 11:10 | <Philip`> | ako: I think the idea is it's something the application can handle itself, and the GL library should be as small as possible and avoid that stuff |
| 11:10 | <Philip`> | (in GLES, that is) |
| 11:10 | <othermaciej> | WebGL is roughly modeled on GLES 2.0 |
| 11:11 | <ako> | Many other areas of functionality have been removed in version 1.0 to produce a lightweight interface: for example, quad and polygon primitive rendering[...], push and pop state attributes, [...]. |
| 11:11 | <ako> | that's what wikipedia says .) |
| 11:11 | <Philip`> | "Client and server attribute stacks are not supported by the profiles; consequently, the commands PushAttrib, PopAttrib, PushClientAttrib, and PopClientAttrib are not supported." |
| 11:12 | <Philip`> | That's what the spec says |
| 11:12 | <ako> | but there are also things like push/popmatrix in opengl |
| 11:14 | <Dashiva> | Yeah, the matrix stacks are kinda important |
| 11:14 | <ako> | ah... there is push/popmatrix in gles 1.0 |
| 11:15 | <gsnedders> | zcorpan: Really, do you need to spend emails that lead to names like the "script data escaped end tag open state"? |
| 11:15 | <ako> | guess that makes sense... all that fixed pipeline stuff was stripped away in gles2.0 |
| 11:15 | <zcorpan> | gsnedders: yes |
| 11:16 | <gsnedders> | zcorpan: Damn you. |
| 11:16 | gsnedders | is currently implementing that one |
| 11:17 | <zcorpan> | so when's WHATWG LC? |
| 11:17 | <Hixie> | probably monday |
| 11:17 | <gsnedders> | As far as I can tell, when Hixie gets around to announcing it |
| 11:17 | <Hixie> | or tuesday |
| 11:17 | <gsnedders> | Hixie: Can you fix the parser bugs I reported? Pretty please? |
| 11:17 | <zcorpan> | cool |
| 11:17 | <Hixie> | gsnedders: yeah, gonna do that in the next hour or so. responding to mail about websockets first. |
| 11:18 | <zcorpan> | Hixie: good work |
| 11:18 | <gsnedders> | Hixie: thx, I guess I'll get the impl done of everything when I get back to Linköping, but likely tomorrow, after sleeping |
| 11:19 | gsnedders | doesn't think allowing himself to be bullied into dancing for four hours last night was a good idea |
| 11:21 | <Hixie> | zcorpan: i'm more impressed by my timeline prediction being accurate |
| 11:21 | <aho> | generally the canvas spec stuff is pretty well done though. eg it does specifiy the clear color (transparent black). java (java2d that is) for example didn't get this right |
| 11:23 | gsnedders | wonders why jgraham was looking at him |
| 11:24 | <Hixie> | oh hey, good news. according to the htmlwg charter (the "realistic" timetable, vs the one i predicted), we'll be at PR in Q2 next year! |
| 11:24 | <Hixie> | and we've been at CR for a years now |
| 11:24 | <Philip`> | aho: If there's anything that it doesn't specify pretty well, please file bugs :-) |
| 11:24 | <Hixie> | (LC was summer last year!) |
| 11:24 | <zcorpan> | Hixie: maybe you just worked your ass off to meet the prediction |
| 11:24 | <Hixie> | i did |
| 11:25 | <Hixie> | for 6 years :-P |
| 11:25 | <Philip`> | You can't stop now :-p |
| 11:25 | <aho> | Philip`, well... save/restore sucks :> (but since it's js it isn't much of a deal... one can just make push/pop point at those functions) |
| 11:26 | <zcorpan> | Hixie: yeah but you've been on a spree this year :) |
| 11:26 | <Philip`> | aho: Well, I meant "well" more in the sense of "precisely", rather than "good design" :-) |
| 11:26 | <Hixie> | zcorpan: no more than previously, i don't think. It's just been more feedback focussed this year rather than new-feature focussed |
| 11:27 | <zcorpan> | Hixie: ok |
| 11:28 | <aho> | Philip`, i'm only interested in a very small subset... but these things do look just fine |
| 11:28 | <aho> | (transformations, images, solid fill, and fillrect - that's it) |
| 11:30 | <aho> | it's the minimum one needs for games... and it's the maximum i'm willing to implement ;) |
| 11:31 | Philip` | wonders what aho is implementing it for |
| 11:31 | <aho> | some small embedded thingy |
| 11:32 | <aho> | v8 is far smaller than python and about 10 times faster... it also starts up far quicker |
| 11:32 | <Philip`> | Ah |
| 11:33 | <aho> | ye, it's also far quicker than bloody as2 (got a rather bugged implementation there) :f |
| 11:34 | <aho> | the canvas subset is so interesting because you can easily test it anywhere |
| 11:34 | <aho> | that's the big idea... basically |
| 11:34 | <aho> | well, and the overhead is rather small |
| 11:36 | <aho> | webgl would have been also great, but unfortunately the device only supports gles 1.1 |
| 11:37 | jgraham | notes that gsnedders is between him and the windows so he was likely not looking at gsnedders at all |
| 11:37 | gsnedders | was guessing that might be the case, but it's true that it's really not interesting what you can see through the window |
| 11:47 | gsnedders | is finally close to the end of all the script states |
| 11:47 | gsnedders | glares at zcorpan |
| 11:48 | <Hixie> | gsnedders: btw in the future if you want me to prioritise a bug, mark it critical, so it goes red in the bug list |
| 11:49 | <Hixie> | is http://www.w3.org/Bugs/Public/show_bug.cgi?id=8019 the one you want me to deal with? |
| 11:49 | <gsnedders> | 8029 most urgently |
| 11:50 | <gsnedders> | Then 8019 just so I can claim the tests/impl actually match the spec |
| 11:50 | <zcorpan> | does someone know what the correct -o-link syntax is? |
| 11:53 | <Philip`> | zcorpan: Probably like http://software.hixie.ch/utilities/js/live-dom-viewer/saved/160 |
| 11:53 | <Hixie> | gsnedders: done 8029 |
| 11:53 | <gsnedders> | Hixie: Buses have arrived |
| 11:53 | <Hixie> | later |
| 12:20 | Philip` | is made happier |
| 12:20 | <Philip`> | though I suppose now I have to read and check the new version :-( |
| 12:50 | <Philip`> | The "set of space-separated tokens" definition seems somewhat pointless, since it's effectively saying a set of space-separated tokens is any string with zero or more characters |
| 12:50 | <Philip`> | so it's just a synonym of "string" |
| 12:51 | <Philip`> | Oh, that's not true, because " " isn't a set of space-separated tokens, though it's valid when you have a string containing a set of space-separated tokens |
| 12:52 | <Philip`> | but since any set of etc is a string, and certainly contains a set of etc, then presumably the set of etc may have leading and trailing spaces |
| 12:52 | <Philip`> | so it's just a convoluted synonym of "string" |
| 17:16 | gsnedders | looks a logs, and notices he didn't tell Hixie what he meant to earlier |
| 17:16 | <gsnedders> | s/Buses have arrived/I'll look later, as the buses have arrived/ |
| 22:37 | <deadowl> | Is there a continuity tag proposal yet? |
| 22:42 | <deadowl> | Hi Rimdeker |
| 22:42 | <Rimdeker> | Hi deadbowl |
| 22:42 | <Rimdeker> | lol |
| 22:42 | <Rimdeker> | How are you? Long time no see. |
| 22:42 | <deadowl> | I'm okay |
| 22:43 | <Rimdeker> | What did you do with your share of THE money, you know what I'm talking about? |
| 22:43 | <deadowl> | I want to know whether there's going to be a continuity tag in HTML 5. |
| 22:43 | <deadowl> | I spent it on a futon because my bed imploded. |
| 22:43 | <Rimdeker> | ...that's gotta be one damn expensive futon |
| 22:44 | <deadowl> | well, the frame is made out of the bones of notable mathematicians throughout history. |
| 22:44 | <deadowl> | so that was kind of expensive to make happen. |
| 22:45 | <Rimdeker> | Disturbing yet interesting. |
| 22:45 | <deadowl> | Do you know what I'm talking about by continuity tag?<thereismore>...</thereismore> |
| 22:45 | <Rimdeker> | I'll be honest with you, I started learning HTML , CSS and PHP about a week ago |
| 22:45 | <Rimdeker> | I know nothign |
| 22:46 | <deadowl> | I learned html 10 years ago, css like 8 years ago, and php last year. |
| 22:46 | <Rimdeker> | Somehow you make it sound like a painful expierience |
| 22:47 | <Rimdeker> | http://rimdeker.org is technically my first ever website, really do not be fooled by the awesome looks lol, I totally suck in webcoding |
| 22:47 | <deadowl> | oh, it's only painful if you try to maintain an expanding website and don't know a serverside programming language. |
| 22:48 | <deadowl> | of course when I was 12 and learning HTML, I didn't have a clue what a serverside programming language was. |
| 22:49 | <deadowl> | and xhtml/dhtml sounded intimidating at that point in my life. |
| 22:49 | <deadowl> | my mind is more spongey these days somehow |
| 22:49 | <Rimdeker> | ...basically you expanded and maintained websites with html only for almost a decade? |
| 22:49 | gsnedders | wonders how old he was when he learnt HTML now… |
| 22:50 | <deadowl> | Rimdeker: no, I used frames and didn't make complex websites. |
| 22:51 | <deadowl> | It's like... okay, a FAQ site. |
| 22:51 | gsnedders | guesses he started playing around with HTML a bit when he was 8 or 9 |
| 22:51 | <deadowl> | I wish I had a computer when I was 8 or 9. |
| 22:52 | <deadowl> | I got my mom to defect before peak aol. |
| 22:52 | <gsnedders> | And I must've been 11 when I finally got bullied into standards and this HTML5-like thing that'll replace HTML called XHTML. |
| 22:52 | <deadowl> | xhtml is html5-like? |
| 22:53 | <gsnedders> | deadowl: Well, this predates HTML 5 existing, and if it is the replacement of HTML 4.01, then it is sort of :P |
| 22:53 | <Rimdeker> | Isn't it actually other way around? |
| 22:53 | <Hixie> | gsnedders: re http://www.w3.org/Bugs/Public/show_bug.cgi?id=8019 |
| 22:53 | <gsnedders> | 14 when I joined WHATWG mailing list I guess |
| 22:53 | <Rimdeker> | I thought HTML 5 is gonna combine HTML and XHTML |
| 22:53 | <Hixie> | gsnedders: there are two errors -- missing space, and missing quote |
| 22:54 | <Hixie> | gsnedders: you want just one reported? |
| 22:54 | <gsnedders> | Hixie: Yeah, I don't think it makes sense to throw multiple parse errors for a single character |
| 22:54 | <Hixie> | k |
| 22:54 | <gsnedders> | (and I don't want to update the html5lib tests again) :P |
| 22:55 | <Rimdeker> | So Ian, how are you today? |
| 22:55 | <Rimdeker> | Wie ist es so in der Schweiz) |
| 22:55 | <Rimdeker> | ?* |
| 22:55 | <Hixie> | actually i am from la suisse romande |
| 22:55 | <Hixie> | and i live in california at the moment :-) |
| 22:56 | <Rimdeker> | ...I suppose ian⊙hc is not you and I suppose your IRC information saying Norway is not true either |
| 22:56 | <Hixie> | actually it is me |
| 22:56 | <Hixie> | and my irc client is in norway |
| 22:56 | <Hixie> | i just connect to it using ssh from california |
| 22:57 | <gsnedders> | Oh duh, 8018 really is dumb |
| 22:57 | <Rimdeker> | Ian , I like pie. |
| 22:57 | <Rimdeker> | That said what's HTML 5 gonna be like? |
| 22:57 | <Hixie> | pretty similar to http://whatwg.org/html5, i guess |
| 22:58 | <gsnedders> | Rimdeker: People called XHTML 1.0 "HTML 5" when the W3C was pushing XHTML as a replacement for HTML several years back. |
| 22:58 | <deadowl> | Okay, so I was wondering is if there is a tag that can be used to express a disconnection in continuity |
| 22:58 | <gsnedders> | deadowl: section? aside? |
| 22:58 | <gsnedders> | Hixie: Are there any other cases in the parser where a single char can throw multiple parse errors? |
| 22:59 | <deadowl> | gsnedders: This is a sentence and<disconnect>...</disconnect> |
| 22:59 | <gsnedders> | deadowl: Also, why do you want to mark that up? What do you gain by marking it up? |
| 23:00 | <deadowl> | gsnedders: it would be useful in contexts that aren't necessarily well-formed |
| 23:00 | <Hixie> | gsnedders: i hope not, but Philip` might know |
| 23:00 | <deadowl> | gsnedders: like DOM Ranges |
| 23:01 | <gsnedders> | deadowl: I guess I'm missing what you're trying to do |
| 23:01 | <gsnedders> | Hixie: If you hope there aren't, I guess this case shouldn't be there either :) |
| 23:01 | <deadowl> | gsnedders: let's say you have an unordered list, you know how document.selection can go part way into a list? |
| 23:02 | <gsnedders> | deadowl: yeah |
| 23:02 | <Hixie> | gsnedders: you already convinced me dude :-) |
| 23:02 | <gsnedders> | Hixie: :P |
| 23:03 | <gsnedders> | Anyhow, I scarcely got any sleep last night after party/disco at Eng. seminar, so I kinda need sleep |
| 23:03 | <Hixie> | though for sanity my fix is a little different than your request |
| 23:03 | <Philip`> | Hixie: I don't know! |
| 23:03 | <Hixie> | Philip`: you have ocaml magic |
| 23:03 | Philip` | doesn't like being accused of knowing things :-( |
| 23:03 | <Philip`> | Hixie: It's years out of date, though |
| 23:03 | <gsnedders> | Hixie: I just threw together the first thing I could think of that would fix it, I didn't claim it was elegant :P |
| 23:04 | <Hixie> | Philip`: didn't you have some regexps to nearly auto-generate it? |
| 23:04 | <gsnedders> | Hixie: Is there anything that can implemented in an hour or less than you want in Anolis soon? |
| 23:04 | <Philip`> | Hixie: Yes, except not very "auto", since the spec used lots of inconsistent phrasing so it needed hundreds of regexps to match everything |
| 23:05 | <Philip`> | and it couldn't automatically deduce high-level meanings, like phases or modes or whatever they are |
| 23:05 | <Philip`> | so that structure had to be hard-coded |
| 23:05 | <Philip`> | and then you changed it all :-( |
| 23:05 | <gsnedders> | (Equally, that question about Anolis goes to anyone else who wants anything soon) |
| 23:06 | <deadowl> | gsnedders: if I put a range over something that's well formed, and then extract or clone it's contents. I have no way of knowing in that range whether or not the well-formed contents of the clone/extraction reflect the well-formed origin. Hence the utility of having a tag to signify a lack of continuity. |
| 23:07 | <gsnedders> | deadowl: In what way does just comparing nodes not work? |
| 23:07 | <Philip`> | (Also I never finished a complete implementation of the tree construction stuff) |
| 23:07 | <gsnedders> | Anyhow, time for me to deal with four hours of non-stop dancing. Feel free to leave messages to me about Anolis. |
| 23:08 | <gsnedders> | (That's not strictly true, I had a five minute or so break every hour.) |
| 23:08 | <Philip`> | It'd be possible (and possibly useful) to update and fix all the code, but I guess it'd take several days' work |
| 23:08 | <deadowl> | gsnedders: let's say you're doing an XMLHttpRequest for an incomplete list |
| 23:09 | <deadowl> | so you could do something like <ul><li>a</li><li>b</li><li>c</li><disconnect>more...</disconnect></ul> |
| 23:10 | <deadowl> | and then replace the disconnect with a continuity for the list. |
| 23:10 | <deadowl> | that could be followed by another disconnect. |
| 23:10 | <deadowl> | though for me its useful context would be in document selections. |
| 23:10 | <Hixie> | Philip`: heh. |
| 23:10 | <Hixie> | Philip`: it should be more consistent now fwiw |
| 23:11 | <Hixie> | gsnedders i think i'm out of requests for anolis at this point (don't need even cross-doc-cross-refs anymore) |
| 23:12 | Hixie | really should rewrite his sawzall implementation of the parser |
| 23:12 | <Philip`> | Hmm, why was I talking about the tree construction stuff? |
| 23:12 | <Philip`> | The doctype bits are just in the tokeniser |
| 23:13 | <Philip`> | which did (at one point) have a complete working implementation |
| 23:13 | <Philip`> | (though it wasn't generated from regexps, it was just a manual transcription of the algorithm) |
| 23:13 | <Hixie> | it's not that hard to just manually implement the parser from the spec |
| 23:13 | <Hixie> | just tedious |
| 23:14 | <deadowl> | gsnedders: the lexical equivalent for the English language would be ellipses. |
| 23:15 | <Philip`> | ((A manual transcription into a data structure that could then be processed by OCaml code, to make graphs and test cases and to generate implementations, to be specific)) |
| 23:16 | <deadowl> | gsnedders:??? |
| 23:16 | Philip` | wouldn't want to simply write code that implements a parser, because that'd be far too obvious and dull and sensible |
| 23:18 | <Hixie> | that'd be a lot easier than an actual parser ;-) |
| 23:18 | <Hixie> | though it was quite useful iirc |
| 23:18 | <Hixie> | it found a bunch of problems, no? |
| 23:19 | <deadowl> | The fact that nodes can be partially selected in HTML, just as words can be partially selected in English, leaves a gaping hole over the fact that HTML does not have the expressive power to perform the functionality that would be the equivalent of the ellipses in the English language. |
| 23:19 | <deadowl> | anyone? |
| 23:19 | <Hixie> | deadowl: what's the use case? |
| 23:20 | <Hixie> | deadowl: i mean, what kind of site would you use this in and why? |
| 23:20 | <deadowl> | A search engine that contains partial text from a page for instance. |
| 23:21 | <jgraham> | gsnedders: html5lib 0.12 support |
| 23:21 | <deadowl> | Any website that employs pagination. |
| 23:21 | <jgraham> | (also, by extension, html5lib 0.12 to be released) |
| 23:22 | <deadowl> | if i controlled DOM ranges, partially selected nodes in DOM ranges during extraction, etc. |
| 23:22 | <jgraham> | (Mybe we shold call it html5lib 1.0LC to confuse people) |
| 23:23 | <deadowl> | why? so I can lexically distinguish complete data from incomplete data. |
| 23:23 | <deadowl> | should I expand on the use cases more? |
| 23:30 | <Philip`> | Since when did select.value work? I thought I still had to do select.options[select.selectedIndex].value |
| 23:33 | <deadowl> | Hixie: abstract use case: http://pastebin.org/48025 |
| 23:34 | <Hixie> | deadowl: i don't understand what problem the "break" there is helping with that you couldn't do with <li class="more">...</li> |
| 23:35 | <deadowl> | Hixie: it's the transformation that's more important, and being able to reassemble it from across contexts with limited information. Currently DOM ranges don't add an <li class="more"> tag when they extract partially selected nodes. |
| 23:36 | <Hixie> | deadowl: then your use case didn't seem to really show what you meant :-) |
| 23:36 | <deadowl> | it's not only lexically meaningful, it's generalizable to the DOM core. |
| 23:36 | <Hixie> | deadowl: could you show a concrete use case showing what problem you think this fixes? |
| 23:36 | <Hixie> | i.e. what code would look like today that is a problem? |
| 23:37 | <deadowl> | Hixie: you mean the code where I added the requirement for well formed XML? |
| 23:37 | <Hixie> | i don't know |
| 23:37 | <Hixie> | i don't understand what problem we are trying to solve with this |
| 23:38 | <deadowl> | well, in any case, partially it's annoyance with not being able to serverside DOM interpret a file that contains partial html. |
| 23:38 | <deadowl> | it's very much an ajax communication issue. |
| 23:39 | <deadowl> | oh, another use case actually |
| 23:39 | <deadowl> | XHR readyState=3 |
| 23:39 | <deadowl> | where responseText is loaded but responseXML isn't because of the well-formedness requirement. |
| 23:40 | <deadowl> | you could partially load the responseXML if you had a tag to specify where continuity left off. |
| 23:40 | <Hixie> | so the problem is "you can't read a partial responseXML when there is a well-formedness error in the XHR data"? |
| 23:41 | <deadowl> | that wasn't technically the use case that brought it to mind, but it was one that I just thought of |
| 23:41 | <Hixie> | i don't understand the other problem... could you phrase it in the way i just phrased the xhr one? |
| 23:42 | <deadowl> | Hixie: One problem is that response.readyState==3 isn't useful if you're working with XML data. The XML data can't be accessed until request.readyState==4, even if a usable amount of XML is already loaded. |
| 23:44 | <deadowl> | a continuity tag would allow you to easily reconnect multiple pieces of XML broken up into smaller parts if used at DOM level. |
| 23:44 | <Philip`> | Hixie: Why doesn't http://www.w3.org/Bugs/Public/show_bug.cgi?id=8023 have a diff? |
| 23:44 | <deadowl> | At document level it could be used to expand paginated data. |
| 23:44 | <deadowl> | Oh, it could also be used to connect diffs. |
| 23:45 | <deadowl> | interchangably even. |
| 23:47 | <Philip`> | deadowl: Do you mean you could have XML like "<a> x <b> y | z </b> w </a>", split it at the "|" into two XML documents with some extra tag, and then merge them back into the original? |
| 23:53 | <Hixie> | Philip`: dunno, script failure somewhere i guess. i added it manually. |
| 23:54 | <Hixie> | deadowl: sounds like it would make life a lot more complicated than just use document fragments or a custom DOM node, or even a range showing what's new |
| 23:54 | <Hixie> | deadowl: generally speaking you're going about this wrong, though -- we prefer to start with a clear problem description, and no idea of the solution, and then to consider solutions by evaluating them against how well they resolve the problem |
| 23:55 | <deadowl> | yea |
| 23:55 | <Philip`> | Hixie: Thanks |
| 23:56 | Philip` | notes that the bug reporting system didn't cope very well with Unicode in that bug |
| 23:56 | <deadowl> | i'm going backward in my mind though because i thought of this as the best solution after having a problem, but its been a while since aving the problem |
| 23:56 | <Hixie> | Philip`: looks fine to me... |
| 23:57 | <Philip`> | http://www.w3.org/Bugs/Public/show_bug.cgi?id=8023 looks like "≤" to me (in the title and comment text) |
| 23:58 | <Hixie> | maybe i set my default to utf-8 |
| 23:58 | Philip` | gets the problem in Opera and Firefox |