| 00:01 | <takkaria> | Hixie: not the dreaded meetings? I hope they're not regular ones |
| 01:08 | <tomg> | I stupidly wrote an email to whatwg⊙wo before I'd confirmed my list membership, will it eventually get approved? :) |
| 01:08 | <tomg> | or be lost forever in a rampant jungle of spam |
| 01:09 | <Philip`> | tomg: Things in the moderation queue never get approved |
| 01:14 | <tomg> | I shall resend |
| 01:14 | <tomg> | ta |
| 05:14 | Hixie | steels himself for a flaming |
| 05:16 | <hober> | oh? |
| 05:17 | <Hixie> | the js list |
| 05:19 | <hober> | ahh. that's in my other emacs. looking... |
| 05:31 | <Hixie> | hey i hadn't seen this before http://edge-op.org/iowa/www.iowaconsumercase.org/011607/2000/PX02991.pdf |
| 07:44 | <Hixie> | http://edge-op.org/iowa/www.iowaconsumercase.org/010807/PLEX_5735.pdf is interesting too |
| 07:47 | <Hixie> | http://edge-op.org/iowa/www.iowaconsumercase.org/010807/PLEX_5803.pdf too |
| 07:59 | <Hixie> | http://edge-op.org/iowa/www.iowaconsumercase.org/010807/PLEX_5906.pdf brings up an interesting point |
| 08:00 | <Hixie> | one way to make the Web platform safer from proprietary embrace-and-extend attempts would be to have a certification mechanism that only certifies web pages that don't use anything proprietary |
| 08:00 | <Hixie> | sort of like the "Valid HTML4" buttons, but not specifically for conformance, but for lack of using anything that isn't part of the core cross-platform Web standards |
| 08:01 | <Hixie> | i just talk to hsivonen about this |
| 08:03 | <roc_> | Myhrvold is smart |
| 08:05 | <Hixie> | Myhrvold? |
| 08:06 | <hsivonen> | Hixie: the billg email about Office HTML export is interesting. |
| 08:06 | <hsivonen> | especially considering the colon legacy today |
| 08:07 | <hsivonen> | Hixie: detecting the absence of proprietary stuff programmatically will be a hard game of whack-a-mole |
| 08:07 | <hsivonen> | Hixie: first, checking script API exposure statically is hard |
| 08:08 | <hsivonen> | Hixie: second, whitelisting standard stuff is not enough due to things like conditional comments |
| 08:08 | <Hixie> | it would be hard work, yes |
| 08:08 | <roc_> | Myhrvold sorry |
| 08:09 | <hsivonen> | Hixie: proprietary stuff would appear where version n of the checker wouldn't know to look and n+1 would have to blacklist that stuff |
| 08:09 | <roc_> | he's the primary author of PLEX_5803. The first two paragraphs are Gates, the rest is Myhrvold |
| 08:09 | <Hixie> | ah ok |
| 08:09 | <Hixie> | oh the nathan guy |
| 08:09 | <Hixie> | yes |
| 08:09 | <Hixie> | very |
| 08:09 | <roc_> | not long after that memo he got bored with Microsoft and left to do research on dinosaur tails |
| 08:10 | <Hixie> | seems to have respect issues with other employees |
| 08:10 | <Hixie> | but that's another story |
| 08:10 | <Hixie> | heh |
| 08:11 | <Hixie> | hsivonen: yeah, it would be difficult. might be an interesting thing to persue nonetheless, i think. :-) |
| 08:13 | <hsivonen> | Hixie: I doubt all schemes where software gives a Web author a badge. Software like Validator.nu works when the author uses it in order to solve his development issues privately. |
| 08:13 | <hsivonen> | but when you start giving out badges, the motivations of the users get distorted badly |
| 08:13 | <Hixie> | the problem this would be solving has, frankly, little to do with the author's own goals |
| 08:13 | <Hixie> | i agree that gaming the system is a tough one, though |
| 08:14 | <Hixie> | as in, a danger to any such scheme |
| 08:14 | <hsivonen> | consider accessibility checking |
| 08:14 | <Hixie> | it would have to be continuously monitored and maintained |
| 08:14 | <hsivonen> | if you really want to make an accessible site, you need to hire people |
| 08:14 | <Hixie> | automated accessibility checking is a non-starter. |
| 08:14 | <Hixie> | it's like usability checking. hell it _is_ usability checking. |
| 08:15 | <hsivonen> | if you want to give the appearance of having done something in the hope that you don't get sued or in the hope of selling stuff to non-disabled buyers, you game an accessibility checker program |
| 08:16 | <Hixie> | yup |
| 08:29 | <hsivonen> | http://del.icio.us/rogerjohansson/dictatorship |
| 08:32 | <roc_> | ooh, nice ligatures in the header in Firefox 3! |
| 08:34 | <hsivonen> | http://a.viary.com/tools/ they sure are advancing on a lot of fronts simultaneously. too bad it's flash--not canvas |
| 08:36 | <jruderman> | ligatures on http://blog.fawny.org/2007/12/23/janefonda/ ? yeah, easy to see the difference between firefox trunk and safari 3 there |
| 08:36 | <jruderman> | zapfino i'm guessing (domi seems to be broken) |
| 08:39 | <Hixie> | http://edge-op.org/iowa/www.iowaconsumercase.org/122106/PLEX0_2272.pdf is funny because they very bluntly put the Web's openness in the "con" category |
| 08:48 | <Hixie> | http://edge-op.org/iowa/www.iowaconsumercase.org/122106/PLEX0_2278.pdf documents their intent to embrace-then-extend HTML |
| 08:55 | <annevk> | it also says to keep Office formats alive and not allow other formats as that might make it easier for people to use other software |
| 08:55 | <annevk> | (other format cited is HTML) |
| 08:56 | <Hixie> | holy shit, http://edge-op.org/iowa/www.iowaconsumercase.org/122106/PLEX0_2494.pdf suggests putting hte IE code out there to let peopl experiment with it |
| 08:56 | <annevk> | what is this stuff about? |
| 08:56 | <Hixie> | i guess that didn't go down well! |
| 08:56 | <Hixie> | files from Comes v. Microsoft |
| 08:58 | <annevk> | interesting |
| 09:00 | <annevk> | the embrace and extend HTML thingie also mentions Blackbird as other strategy |
| 09:00 | <annevk> | they're trying that one again now... it's called SilverLight |
| 09:00 | <hsivonen> | what was Blackbird? |
| 09:00 | <annevk> | similar thing to SilverLight as I understand it |
| 09:02 | <Hixie> | blackbird was their msn client, i believe |
| 09:02 | <Hixie> | aol competitor |
| 09:02 | <Hixie> | i may be wrong, the google probably knows |
| 09:02 | <Hixie> | http://edge-op.org/iowa/www.iowaconsumercase.org/122106/PLEX0_2658.pdf suggests they knew that IE was killing Windows years before they finally pulled the plug |
| 09:02 | <annevk> | wikipedia says: "Online content authoring technology developed alongside MSN 1.0 but cancelled in favor of HTML and ActiveX" |
| 09:02 | <Hixie> | right |
| 09:03 | <annevk> | from http://en.wikipedia.org/wiki/Microsoft_codenames |
| 09:03 | <hsivonen> | ok |
| 09:04 | <annevk> | "I agree that making sure applicatiosn are primarily on Windows is something we lost ... of" |
| 09:05 | <annevk> | "I rail against the people who want to just give things away like DirectX" |
| 09:06 | <annevk> | "We should have people laughing at the idea of 100% pure Java whether they wnte in JAVA or not" |
| 09:06 | <annevk> | s/wnte/write/ |
| 09:07 | <annevk> | (character recognition is not perfect) |
| 09:15 | <Hixie> | IE4 usability study report http://edge-op.org/iowa/www.iowaconsumercase.org/122106/PLEX0_2701.pdf |
| 09:15 | <Hixie> | interesting to note that they barely changed any of the things that that report lambasts |
| 09:17 | <hsivonen> | how many times does the U.S. legal process fax or photocopy these docs? |
| 09:19 | <Hixie> | yeah i love how these docs started as digital data and got utterly mangled before turning back into digital data |
| 09:20 | <Hixie> | some of these papers are really amusing given what Microsoft is doing with .NET/Silverlight/WPF |
| 09:21 | <Hixie> | like, it's exactly the same thing, just 12 years later and with a different name |
| 09:23 | <annevk> | guess they still hae the same strategy |
| 09:44 | <roc_> | it's not the same thing |
| 09:44 | <roc_> | the funny thing is that Silverlight is Microsoft's best play right now and it's cross platform |
| 09:45 | <roc_> | Jim Allchin must be mad |
| 09:45 | <annevk> | http://simonwillison.net/2008/Mar/11/high/ how does it download scripts in parallel? |
| 09:46 | <annevk> | that shouldn't be possible without <script async> |
| 09:46 | <roc_> | it's not that hard |
| 09:46 | <roc_> | you parse ahead and prefetch things that look like <script> tags |
| 09:46 | <roc_> | we have a bug on it |
| 09:47 | <Philip`> | I guess that'd go wrong when the first script does document.write('<!--') |
| 09:47 | <Camaban> | "The HTTP 1.1 spec recommends that browsers only download two items in parallel per hostname, but the spec was written in 1999. Today?s clients and servers can support more parallel downloads, so IE8 has increased the number of downloads per hostname from 2 to 6." |
| 09:47 | <annevk> | yeah, that'd make parsing non-deterministic |
| 09:47 | <roc_> | no |
| 09:47 | <roc_> | you fetch optimistically |
| 09:48 | <roc_> | you don't actually parse or execute anything until you've finished the earlier script |
| 09:48 | <annevk> | <script src=hideevil.js></script><script src=evil?delete=foo></script> |
| 09:48 | <roc_> | yeah you could do that |
| 09:48 | <roc_> | feel free :-) |
| 09:48 | <Philip`> | Fortunately <script src> does GET and GET doesn't have side effects |
| 09:49 | <annevk> | hah |
| 09:49 | <roc_> | yeah right |
| 09:49 | <roc_> | BTW Blackbird was back from when MSN was building a "walled garden" like AOL |
| 09:50 | <roc_> | so you'd negotiate with Microsoft to get your Blackbird app installed in the MSN garden |
| 09:50 | <roc_> | I knew a guy at CMU who had a contract with Microsoft to get the Internet Chess Club into MSN |
| 09:50 | <roc_> | then MS saw the writing on the wall, cancelled the entire idea and tore up all the contracts |
| 09:51 | Philip` | remembers a neat 3D chat client on the old MSN |
| 09:51 | <Philip`> | but I don't remember anything more than its existence :-( |
| 09:53 | hsivonen | has never been on AOL or MSN |
| 09:53 | <hsivonen> | or Minitel for that matter |
| 09:54 | <hsivonen> | the Web is larger that any country-focused walled garden |
| 09:54 | <Philip`> | The web was much more boring back then |
| 09:59 | <Hixie> | http://edge-op.org/iowa/www.iowaconsumercase.org/122106/PLEX0_5879.pdf is interesting, for those who were interested in the office/IE link earlier |
| 09:59 | <Hixie> | roc_: how is it not the same thing? |
| 10:00 | <Hixie> | roc_: silverlight seems almost identical to their reaction to java, as well as their reaction to netscape |
| 10:00 | <roc_> | IE ended up being Windows only |
| 10:00 | <Hixie> | roc_: create a platform, put it on all the other platforms (mac and unix), and then pull the rug out of those when you have enough of the market |
| 10:00 | <annevk> | that stuff is great Hixie |
| 10:00 | <roc_> | IE was never really on Mac and Unix |
| 10:00 | <Hixie> | silverlight will be windows only too |
| 10:01 | <Hixie> | neither is WPF now |
| 10:01 | <roc_> | I'm not a big Silverlight fan, as you may be aware |
| 10:02 | <Hixie> | who is :-) |
| 10:02 | <Hixie> | apart from miguel, i mean |
| 10:02 | <Camaban> | my boss :| |
| 10:02 | <roc_> | but Silverlight looks a lot more cross-platform than IE ever did |
| 10:02 | <roc_> | a big part of the IE story was ActiveX |
| 10:02 | <annevk> | i thought some open source guy ported SilverLight |
| 10:03 | <annevk> | (related to mono maybe?) |
| 10:03 | <roc_> | annevk: are you joking with us? |
| 10:03 | <annevk> | iirc it is sort of cross-platform though |
| 10:03 | <Hixie> | roc_: silverlight is the tiny part of WPF in the same way that HTML was the tiny part of IE |
| 10:03 | <roc_> | I don't think so |
| 10:03 | <roc_> | Silverlight is the small, auto-updating download |
| 10:03 | <annevk> | http://www.mono-project.com/news/archive/2007/May-05.html |
| 10:03 | <roc_> | WPF is huge, requires .NET 3 |
| 10:04 | <roc_> | and is slow and bloated according to the MS shippingseven blogger |
| 10:04 | <Hixie> | well i'm not saying their execution is any good |
| 10:04 | <Hixie> | i'm just describing the theory |
| 10:04 | <roc_> | I agree there are multiple ways they could leverage Silverlight in the future |
| 10:05 | <roc_> | but right now it does not protect Windows |
| 10:05 | <Hixie> | silverlight is "WPF/E" |
| 10:05 | <Hixie> | sure |
| 10:05 | <hsivonen> | so is the next step for Adobe to put Flash Player under MPL/GPL/LGPL? |
| 10:05 | <roc_> | silverlight is done by a different team BTW |
| 10:05 | <Hixie> | just like IE doesn't protect windows |
| 10:06 | <roc_> | IE does protect Windows because MS never shipped a compatible IE on Mac |
| 10:06 | <Hixie> | roc_: different team, or different division? because each opera device is done by a different team, for example, doesn't stop them being the same product :-) |
| 10:06 | <hsivonen> | given the Flash legacy out there, I'd like to have Gecko-native Flash support with nice integration with Cairo and the event loop |
| 10:06 | <Philip`> | hsivonen: About the only time I use Flash content is for watching videos, and I guess it's quite unlikely for the video codecs to be released as MPL/GPL/LGPL |
| 10:07 | <Hixie> | roc_: from these papers, it doesn't seem like they considered IE/Mac an incompatible alternative, they always refer to it more as a subset |
| 10:07 | <roc_> | it ended up being an incompatible alternative |
| 10:07 | <Philip`> | (I've heard Moonlight has the same problem with video, and relies on a binary component from Microsoft) |
| 10:07 | <roc_> | one of these memos said "why are we doing IE4 cross platform! stop that!" and they did |
| 10:08 | <roc_> | Philip`: that is true |
| 10:08 | <hsivonen> | Philip`: Flash is used awfully lot on sites that try to create a brand image |
| 10:08 | <hsivonen> | Philip`: it's crazy how much it is used on sites of clothing brands |
| 10:08 | <hsivonen> | even though a big use case for the sites of clothing sites is using a store locator when you are on the move |
| 10:08 | <roc_> | Hixie: the environment currently demands that a credible Internet platform be cross-platform |
| 10:08 | <roc_> | but two things have changed since the mid 90s |
| 10:09 | <roc_> | Apple is no longer going down the toilet |
| 10:09 | <hsivonen> | I remember a case when my mother called me from abroad and asked me to act as a voice interface to the Web in order to locate a store |
| 10:09 | <roc_> | and there's a big market of phones that Microsoft doesn't control |
| 10:09 | <hsivonen> | the relevant site used flash in a bad way |
| 10:09 | <hsivonen> | and I had to try three browsers until I was able to extract the information |
| 10:10 | <Hixie> | roc_: like i said, i'm not saying their execution is any good :-) |
| 10:10 | <Hixie> | roc_: it does still seem to me like they're just going through the same motions again though |
| 10:12 | <hsivonen> | Philip`: anyway, in order to dig into a store locator from a browsing setup that Adobe hasn't taken care of, integrated Open Source Flash would be good even without patented codecs |
| 10:12 | <roc_> | I think the same thinking is going on internally, but conditions are different, and not as favourable to MS |
| 10:13 | <roc_> | I don't see that they will be able to compromise on the cross-platform-ness of Silverlight any time soon |
| 10:13 | <roc_> | and if we and Adobe don't drop the ball, we can keep pushing that out |
| 10:14 | <roc_> | that leaves Windows unprotected indefinitely |
| 10:18 | <Hixie> | yeah, i agree that it's not a good position for them this time |
| 12:51 | <Philip`> | It's odd how http://online.wsj.com/public/page/election2008.html sends the request headers back as part of its response |
| 14:01 | <hendry> | http://test:test⊙bwn/ # this is probably a http issue, but i had this crazy notion earlier that these types or urls were deprecated. I guess i was wrong or? |
| 14:14 | <annevk> | they are |
| 14:20 | <hendry> | annevk: is there a reference to that somewhere? |
| 14:22 | <annevk> | RFC3896? |
| 14:22 | <annevk> | http://annevankesteren.nl/2006/02/alt#comment-6505 |
| 14:22 | <annevk> | about showing alt= in a tooltip ^^ |
| 14:24 | <hendry> | RFC3896 = Definitions of Managed Objects for the DS3/E3 Interface Type ? |
| 14:26 | <annevk> | 3986 |
| 14:35 | <hendry> | 3.2.1. User Information # isn't the clearest passage I've ever seen. So hashes or implementations not showing the password are allowed? Bit confused. |
| 14:36 | <annevk> | 'Use of the format "user:password" in the userinfo field is deprecated.' |
| 14:43 | <annevk> | roc_, if you still think ClientRect.height/width is a good idea let me know and I'll add them, they probably wouldn't make Opera 9.5 though |
| 15:03 | <Philip`> | Hmm, my regexp-based spec-to-OCaml compiler is not absolutely happy when the spec is all rearranged |
| 15:09 | Philip` | wonders why people sound optimistic about XBL2, when it seems like there's approximately no activity in implementing it |
| 15:12 | <krijnh> | http://code.google.com/p/xbl/ no? |
| 15:13 | <Philip`> | http://weblogs.mozillazine.org/roc/archives/2008/03/not_a_crossbrow.html says no |
| 15:14 | <krijnh> | Ah, what does he know about it |
| 15:14 | <krijnh> | *runs* |
| 15:34 | <Philip`> | From <meta> parsing: "change the encoding to the encoding encoding" sounds funny |
| 16:20 | <hsivonen> | Sam is right about the tiredness of the trajectory |
| 16:20 | <hsivonen> | too bad this issue is closely related to a rathole topic |
| 16:31 | <Philip`> | gsnedders: http://diveintomark.org/archives/2008/03/09/no-fury-like-dracon-scorned#comment-11446 looks a bit off |
| 16:31 | <gsnedders> | Philip`: isn't that the UTF-8 for U+FFFF? |
| 16:32 | <gsnedders> | yeah, it is. sod. |
| 16:32 | <Philip`> | gsnedders: I mean the bits like "That is: “?>." and "Well, let’s have a completely valid line of PHP: ." |
| 16:33 | <Philip`> | I love blogs that don't have a comment preview and/or don't tell you what the comment syntax is |
| 16:33 | <gsnedders> | ah. |
| 16:36 | <Philip`> | (U+FFFF is %ef%bf%bf in UTF-8, and it's the only UTF-8 sequence I've ever memorised, but I can't see what your question is referring to) |
| 16:38 | <gsnedders> | Philip`: well, if he doesn't even allow trackbacks I'm not particularly fused about clarifying it |
| 18:11 | <hsivonen> | http://blog.vlad1.com/2008/03/10/silverwhat/ |
| 18:28 | <hsivonen> | did the correctness flag just change to force quirks with true/false flipped or was there something else? |
| 18:51 | <Dashiva> | hsivonen: say what? |
| 18:53 | <annevk> | hsivonen, it was just a renaming afaict |
| 18:54 | <annevk> | hsivonen, though some small changes have been made to the tokenizer |
| 18:57 | <hsivonen> | annevk: ok. thanks |
| 18:57 | <hsivonen> | Dashiva: the tokenizer changed |
| 19:09 | <hsivonen> | http://annevankesteren.nl/2006/11/dot-mobi#comment-6506 |
| 19:18 | <Hixie> | man, people start debating solutions really quickly |
| 19:18 | <Hixie> | there's been like one or two e-mails on the problems solved by namespaced content in text/html tops |
| 19:19 | <Hixie> | and ironically, most of those are likely best solved by solutions that nobody has even considered |
| 19:19 | <Hixie> | e.g. using <script> and changing the syntax of SVG so that angle brackets use square brackets instead |
| 19:21 | <hsivonen> | Hixie: I should resist responding better even when questions are directed at me :-( |
| 19:21 | <Hixie> | no, your responses are very useful |
| 19:21 | <Hixie> | as are their own e-mails, by and large |
| 19:21 | <hsivonen> | Hixie: anyway, the thread might get back on track if you responded to Sam's last email in the thread |
| 19:21 | <Hixie> | cool, will see if i can respond when i get to it |
| 19:22 | <Hixie> | (today, i mean) |
| 19:22 | <hsivonen> | cool |
| 19:24 | <Hixie> | not clear how to reply |
| 19:26 | <hsivonen> | aren't there any open technical issues needing answers? |
| 19:26 | <hsivonen> | or research items, rather. |
| 19:27 | <Hixie> | the primary problem i have is that the only use cases i'm aware of are the ones you listed |
| 19:27 | <Hixie> | while i'm sure there must be other use cases, because if those are the only ones, then the solution i come up with is going to look a hell of a lot different than what people are talking about |
| 19:28 | <hsivonen> | that looks like a way to reply :-) |
| 19:30 | <hsivonen> | Hixie: I fail to see how you can get to a hell of a lot different if you take the constraint of being able to paste source created in an SVG editor and the requirement to have the same kind of DOM as in the XHTML+SVG case |
| 19:31 | <Hixie> | well, maybe no that different |
| 19:32 | <hsivonen> | somehow I remember flipping the boolean of the doctype token to the exact opposite once before... |
| 19:33 | <Hixie> | m? |
| 19:34 | <hsivonen> | the boolean value should be the opposite when modeling force quirks compared to modeling correctness |
| 19:35 | <Hixie> | oh, you're using a boolean for that |
| 19:35 | <hsivonen> | perhaps I should just call it dontForceQuirks to avoid breaking the interface |
| 19:35 | <Hixie> | i would have used constants other than "true" and "false" :-) |
| 19:37 | <Philip`> | The tokeniser test JSON files have true/false meaning "don't force quirks" |
| 19:37 | <Hixie> | ah |
| 19:37 | <hsivonen> | Philip`: that probably dates from before the previous flip... |
| 19:38 | <hsivonen> | Hixie: when you have a flag that has states "on" and "off", implementing it as a boolean is not that far-fetched... |
| 19:39 | <Hixie> | sure :-) |
| 19:39 | <Hixie> | i usually prefer using constants that map to true and false to make it clearer what they mean in argument calls |
| 19:39 | <Hixie> | so instead of foo(true, true, false) you have foo(doThis, doThat, doTheOther) |
| 19:40 | <Hixie> | question about inkscape. Does it always output un-prefixed content? |
| 19:40 | <Philip`> | setDoctypeCorrectness(true) seems clear enough to me |
| 19:40 | <Philip`> | (I tend not to call my functions foo :-p ) |
| 19:41 | <hsivonen> | Hixie: leopard broke my inkscape, so I can't test, but IIRC, yes at least for docs started in Inkscape |
| 19:41 | <hsivonen> | Hixie: for SVG elements, that is |
| 19:41 | <Hixie> | k |
| 19:42 | <hsivonen> | Hixie: inkscape's own cruft (that renderers are supposed to ignore) has colons |
| 19:43 | <hsivonen> | Hixie: fwiw, the documentation about its own cruft on the inkscape wiki is incomplete |
| 19:43 | <Hixie> | shocking |
| 19:44 | <hsivonen> | considering the inkscape cruft and the DOM fragment serialization idea I mentioned, it would be desirable to put sodipodi: and inkscape: stuff in the right namespaces... |
| 19:44 | <hsivonen> | which kinda opens the door for xmlns... |
| 19:45 | <hsivonen> | then there's the embedded RDF stuff... |
| 19:45 | <Hixie> | we can't use xmlns="" itself |
| 19:45 | <Hixie> | so any solution that ends up going towards a geeric solution loses copy/paste. |
| 19:45 | <hsivonen> | Hixie: we could if it happens only in SVG scope |
| 19:45 | <Hixie> | not in any sane way that is compatible with IE |
| 19:45 | <hsivonen> | :-( |
| 19:46 | <Hixie> | btw what is it about this use case that would exclude using MathML and application/xhtml+xml?: |
| 19:46 | <hsivonen> | anyway, when I uttered inkscape, I wasn't thinking about its app-specific cruft |
| 19:46 | <Hixie> | * Publishing the kind of content that is published on |
| 19:46 | <Hixie> | http://golem.ph.utexas.edu/~distler/blog/ using a |
| 19:46 | <Hixie> | legacy PHP content management system that is not XML-ready. |
| 19:46 | <hsivonen> | the "legacy PHP content management system that is not XML-ready" bit |
| 19:47 | <Hixie> | oh you just mean that you can't be affected by ysod? |
| 19:47 | <Hixie> | ok |
| 19:48 | <hsivonen> | Hixie: yeah |
| 19:49 | <hsivonen> | Hixie: I gather the only way for Jacques Distler's colleagues to publish MathML is by using a system carefully patched by Distler with him on the ysod firefight alert |
| 19:49 | <Hixie> | makes sense |
| 19:50 | <hsivonen> | btw, you might also look into copypasteability from Illustrator |
| 19:50 | <hsivonen> | the illustrator output is evil |
| 19:50 | <hsivonen> | it uses the internal subset to define entities for the namespace URIs |
| 19:51 | <Hixie> | that's fine, i have no reason to believe we ever need to look at namespaces here |
| 19:52 | <Hixie> | does LaTeX have vector graphics |
| 19:52 | <Hixie> | ? |
| 19:52 | <Hixie> | or did you just mean mathematics? |
| 19:52 | <hsivonen> | yes |
| 19:52 | <Hixie> | k |
| 19:53 | <hsivonen> | LaTeX has packages for graph theory and stuff that plots vector graphics |
| 19:53 | <hsivonen> | (plus imported PDF images) |
| 19:55 | <othermaciej> | hsivonen: that sounds incredibly evil |
| 19:56 | <hsivonen> | Hixie: see http://golem.ph.utexas.edu/~distler/blog/archives/001475.html for example of LaTeX code and what it might look like using Web languages |
| 19:59 | <Hixie> | wow that page has use cases! |
| 19:59 | Hixie | notes them |
| 20:00 | <Hixie> | you know, none of the use cases you list in http://lists.w3.org/Archives/Public/public-html/2008Mar/0039.html give any reason not to just have a <latex> element |
| 20:02 | <hsivonen> | Hixie: I amend the requirements by saying that you should be able to resize the CSS formatter view box without re-executing a LaTeX program |
| 20:02 | <Hixie> | :-) |
| 20:02 | <Hixie> | that seems like an optimisation issue :-) |
| 20:02 | <DxSadEagle> | hsivonen: is one allowed to re-execute metafont? :p |
| 20:03 | <hsivonen> | Hixie: also, CSS should be able to tweak the color etc. of math bits from CSS |
| 20:03 | <hsivonen> | DxSadEagle: :-) |
| 20:03 | <Hixie> | oh i agree, i'm just saying that that wasn't on your use cases |
| 20:03 | <Hixie> | and if resizing the CSS allows you to rerun the truetype hinting programs, why can't it also rerun latex? |
| 20:04 | <hsivonen> | Hixie: even though MathML doesn't integrate with a CSS formatter perfectly, it sure integrates a lot better than LaTeX |
| 20:04 | <Hixie> | indeed |
| 20:04 | <Philip`> | TrueType hinting is designed to execute in real-time, LaTeX isn't |
| 20:05 | <hsivonen> | I think I've written about this on the whatwg list a couple of years ago |
| 20:05 | <Philip`> | and it's cheating to just say that's merely performance |
| 20:05 | <Hixie> | hsivonen: probably |
| 20:05 | <Hixie> | Philip`: i agree |
| 20:05 | <Hixie> | i'm just being a devil's advocate |
| 20:05 | hsivonen | searches the archives |
| 20:05 | <Philip`> | because if you don't care about performance you could use a web service API that sends the equation to someone who can write it on a piece of paper and scan it in and send it back |
| 20:08 | <Philip`> | LaTeX is more useful if you're writing a whole document in it, so you can define macros for whatever you're writing about - it'd be much more painful if the LaTeX environment was just a single box on the screen, repeated lots of times |
| 20:08 | <hsivonen> | lists.whatwg.org htdig looks broken |
| 20:09 | <Hixie> | yeah it sucks |
| 20:09 | <Hixie> | there are other sites that let you search the archives |
| 20:09 | hsivonen | tries google |
| 20:09 | <Hixie> | do a google search for "whatwg archives" |
| 20:09 | <Hixie> | Philip`: i'm not in any way seriously suggesting embedding latex anywhere |
| 20:10 | <hsivonen> | http://lists.whatwg.org/pipermail/goanet-goanet.org/2005-July/032308.html |
| 20:10 | <hsivonen> | goanet? |
| 20:10 | <Hixie> | i was just illustrating that the use cases as given in that e-mail don't necessarily preclude solutions that are far removed from what people are discussing |
| 20:10 | <Hixie> | o_O |
| 20:10 | <Hixie> | http://lists.goanet.org/listinfo.cgi/goanet-goanet.org |
| 20:10 | <Hixie> | that's what one gets for using shared hosting i suppose |
| 20:10 | <Hixie> | :-) |
| 20:11 | <Hixie> | though why that url ever came to be, i dunno |
| 20:11 | <Philip`> | http://lists.whatwg.org/pipermail/ |
| 20:11 | <Hixie> | ah |
| 20:11 | <Hixie> | heh |
| 20:13 | <hsivonen> | hmm. I'm pretty sure I've sent email explaining embedded LaTeX renderer baseline and linebreaking problems, but I can't find it |
| 20:17 | <hsivonen> | fwiw, about sending math renderer to an outside service: http://golem.ph.utexas.edu/~distler/blog/archives/000106.html |
| 20:18 | <hsivonen> | http://golem.ph.utexas.edu/~distler/blog/archives/000632.html is better |
| 20:19 | <hsivonen> | “There’s no way to get the baselines to automatically line up, and there’s no way to automatically leave the correct amount of horizontal space in the text to accommodate the equations.” |
| 20:22 | <Philip`> | Has someone already suggested Wikipedia as an example that has lots of maths, and could benefit from higher-resolution text and more consistent fonts and fewer resource requests etc? |
| 20:24 | <Hixie> | hsivonen: did you mean your mail dated Tue, 30 May 2006 ? |
| 20:26 | <shepazu> | Philip`, http://mathworld.wolfram.com/ is also a place that could benefit from inline MathML... have they spoken up on the list? |
| 20:27 | <Hixie> | othermaciej: yt? |
| 20:27 | <othermaciej> | Hixie: yeah, sort of |
| 20:27 | <othermaciej> | (in meeting but can reply slightly asynchronously) |
| 20:28 | <Hixie> | k |
| 20:28 | <hsivonen> | Hixie: I don't see mail from 2006-05-30 from me. I do see http://lists.whatwg.org/pipermail/whatwg-whatwg.org/2006-May/006468.html though |
| 20:28 | <Hixie> | that's the one |
| 20:28 | <Hixie> | i guess timezone issues |
| 20:29 | <hsivonen> | which, surprisingly, doesn't explain well how bad it would be |
| 20:40 | <annevk> | i do have some sympathy for allowing RDFa kind of things in HTML |
| 20:40 | <annevk> | it allows communities to do their own stuff with it without having to resort to quirky things (see microformats) |
| 20:41 | <Dashiva> | Outsource the entire mess |
| 20:42 | <Hixie> | how is RDFa less quirky than microformats? |
| 20:44 | <hsivonen> | it's funny how that question is bracketed by tantek joining and leaving |
| 20:45 | <Dashiva> | There's a lesson to be learned here, kids. |
| 20:46 | <annevk> | Hixie, something better than RDFa would be cool |
| 21:00 | Hixie | got his mac to just read a lot of the table feedback while he was showering, and noticed for the first time that Alex actually takes a breath when talking. |
| 21:00 | <Hixie> | that's crazy |
| 21:00 | <Hixie> | cool, but crazy |
| 21:01 | eseidel | suggests Hixie not use his mac in the shower |
| 21:02 | <Hixie> | well it wasn't IN the shower |
| 21:03 | jwalden | would rather this not go further |
| 21:08 | <Dashiva> | Macbook Water |
| 21:12 | <Philip`> | I can't imagine what a MacBook Fire would be, but I think I want one |
| 21:23 | <roc> | Try building Gecko on a Macbook Pro on your lap |
| 21:24 | <Dashiva> | Maybe it would just have a million firewire ports |
| 21:29 | <othermaciej_> | roc: sounds painful |
| 21:30 | <roc> | instead of -j, 'make' should have an option that caps the maximum temperature |
| 21:31 | Philip` | uses 'cpufreq-set -g powersave' to keep the temperature (and fan noise) down, which is quite effective |
| 21:31 | <othermaciej> | I would not dare build Gecko on a MacBook Pro |
| 21:31 | <othermaciej> | building WebKit is bad enough |
| 21:32 | jgraham__ | wonders if he should mention "being able to use existing authoring tools with minimal amounts of effort" as a requirement for maths in text/html |
| 21:33 | <jgraham__> | It may already be covered |
| 21:33 | <Philip`> | jgraham__: What/which existing authoring tools? |
| 21:33 | <roc> | I do all my development on a MacBook Pro, it's actually great if I avoid skin contact |
| 21:33 | <jgraham__> | Philip`: That matters less to me, as long as they exist. |
| 21:34 | <jgraham__> | (in practice I think that requirement means LateX-like syntax or MathML-like syntax) |
| 21:34 | <roc> | and apart from the fact that some combination of gdb, Firefox and sleeping leaks VRAM on 10.4 |
| 21:34 | <Philip`> | jgraham__: Microsoft Office / OpenOffice.org equation editors might be significant too |
| 21:35 | <jgraham__> | Philip`: I believe oo.o can output MathML |
| 21:35 | <roc> | MathML is the native math format for OpenOffice |
| 21:35 | <roc> | for ODF, I should say |
| 21:36 | <jgraham__> | (in practice LaTeX doesn't need an intermediary but MathML does so my real requirement is "no gratuitous syntax changes from MathML") |
| 21:36 | Philip` | notes the same issue applies to SVG, because only crazy people write vector graphics by hand and everyone else uses Flash or Illustrator or Inkscape |
| 21:37 | <hsivonen> | roc: does OOo now really read back MathML instead of sticking StarMath somewhere and reading it back instead? |
| 21:37 | <roc> | I actually don't know |
| 21:37 | <jgraham__> | Philip`: This is why I would be against e.g. <script type="svg"> and replacing angular brackets with square brackets, fore example |
| 21:38 | <roc> | but at least the MathML is there, and SVG too. It's great to be able to just pull that out and render it |
| 21:38 | <hsivonen> | Philip`: Sam writes SVG in vim... |
| 21:39 | <jgraham__> | hsivonen: Sam is abnormal :) |
| 21:40 | <Philip`> | hsivonen: He is crazy people ;-) |
| 21:40 | <Philip`> | (Of course I might be generalising a bit...) |
| 21:41 | <Philip`> | Incidentally, I use SVG export from Graphviz and Gnuplot much more than from direct editing programs |
| 21:42 | <roc> | Quite a few people are using SVG export from Omnigraffle |
| 21:42 | <roc> | Personally I use OpenOffice Draw |
| 21:42 | <Philip`> | hsivonen: OO.o at least keeps both equation formats when saving, like |
| 21:42 | <Philip`> | <math:math xmlns:math="http://www.w3.org/1998/Math/MathML"><math:semantics><math:mrow><math:mrow><math:msup><math:mi>e</math:mi><math:mrow><math:mi>i</math:mi><math:mo math:stretchy="false">π</math:mo>[snip]</math:mrow><math:annotation math:encoding="StarMath 5.0">e^{i %pi}+1=4</math:annotation></math:semantics></math:math> |
| 21:42 | <hsivonen> | last year, I discovered that OmniGraffle Pro SVG export sucked big time |
| 21:42 | <hsivonen> | so I didn't buy pro |
| 21:42 | <roc> | it outputs some gnarly filters that we have to optimize for |
| 21:43 | <hsivonen> | and instead drafted stuff in OmniGraffle, expoted to PNG and redrew over the PNG in inkscape |
| 21:44 | <Philip`> | If I keep both MathML and StarMath, the StarMath takes precedence when loading |
| 21:44 | <hsivonen> | that's dirty if not evil |
| 21:44 | <roc> | if you delete the StarMath, does it load the MathML? |
| 21:44 | <Philip`> | If I delete the StarMath, it loads from the MathML |
| 21:44 | <roc> | ok cool |
| 21:44 | <Philip`> | and says "{e^{i π} + 1} = 4" |
| 21:45 | <Philip`> | which is about reasonable, except it'd be easier to edit if it said %pi |
| 21:45 | <roc> | it'd be more reasonable if the equation was correct :-) |
| 21:46 | <Philip`> | (Incidentally, the content.xml file says <!DOCTYPE math:math PUBLIC "-//OpenOffice.org//DTD Modified W3C MathML 1.01//EN" "math.dtd"> - I don't know how that differs from normal MathML) |
| 21:46 | <Philip`> | roc: Pfah, that's just constant factors |
| 21:50 | <jgraham__> | Hmm. When I tested OO.o also seemed to have some sort of binary cache for the math object which I had to delete. But I guess I could have been mistaken. |
| 21:51 | <Philip`> | jgraham__: When I just save from OpenOffice.org Math (i.e. not part of a bigger document), it only has XML files and a thumbnail.png |
| 21:52 | <jgraham__> | Ah, I was saving as part of a bigger document |
| 21:52 | <Philip`> | Yay XML - <config:config-item config:name="PrinterSetup" config:type="base64Binary">VAH+/0dlbmVyaWMgUHJpbnRlcgAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAU0dFTlBSVAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAWAAMAmgAAAAAAAAABAAhSAAAEdAAASm9iRGF0YSAxCnByaW50ZXI9R2VuZXJpYyBQcmludGVyCm9yaWVudGF0aW9uPVBvcnRyYWl0CmNvcGllcz0xCm1hcmdpbmRhanVzdG1lbnQ9MCwwLDAsMApjb2xv |
| 21:55 | <Philip`> | When I save from Writer, it has an ObjectReplacments directory containing a binary file for the equation |
| 21:55 | <Philip`> | s//e/ |
| 21:56 | <Philip`> | (...in addition to the standard XML version of it) |
| 21:56 | <jgraham__> | Yeah, that's what I saw |
| 21:58 | <annevk> | It's pretty hard to come up with use cases for "generic extensibility" |
| 21:59 | <annevk> | * metadata annotations (internal format expressed in HTML) |
| 21:59 | <annevk> | * game objects (cf. voidwars.com) |
| 22:00 | <annevk> | maybe in applications it woud sometimes be easier to mint your own elements too |
| 22:00 | <Philip`> | Come up with use cases for specific extensibility, and show that the generic extensibility solution is the 'best' way to handle those cases |
| 22:00 | <annevk> | I try not to think of solutions, but it's hard |
| 22:01 | <jgraham__> | One example of specific extensibility that could, in principle be interesting is distributed references/citations |
| 22:01 | <annevk> | XBL sort of solves the component story, but does something that's optional makes sense if you bind it to <div class="cart"/> |
| 22:01 | <annevk> | or should it be <cart/> |
| 22:02 | <annevk> | (same goes for <planet/> in voidwars fwiw) |
| 22:02 | <jgraham__> | OTOH I don't know if that would actually happen even if the mechanism to make it happen was there |
| 22:03 | <jgraham__> | I've also heard biologists express an interest in embedding machine-readable data in HTML but I forget the details |
| 22:03 | <annevk> | oh, ChemML |
| 22:04 | <annevk> | i wonder if there's a chance that those vocabularies have been hold back because there hasn't been a good way to implement them on top of browsers so far |
| 22:05 | <annevk> | if you have XBL, SVG, and <canvas> you can probably implement most languages as a simple library |
| 22:05 | <jgraham__> | Oh yeah, the Royal Society for Chemistry were doing something that embedded chemistry-specific data in web pages |
| 22:05 | <Philip`> | annevk: But we don't have XBL :-p |
| 22:06 | <annevk> | agreed that XBL is not really realistic right now, but i want it to be! |
| 22:08 | <Philip`> | If it requires all browsers to implement support for it, it seems unlikely it'll ever be realistic |
| 22:08 | <annevk> | why's that? |
| 22:08 | <Philip`> | Because browsers tend to not implement features |
| 22:09 | <roc> | we *want* to implement XBL2 |
| 22:09 | <Philip`> | and nobody would use it until it works adequately in IE |
| 22:10 | Philip` | wants to do lots of things that he'll probably never get around to doing :-( |
| 22:16 | <othermaciej> | all major browsers seem to be on a decent clip of adding features right now |
| 22:17 | <Philip`> | Are any of those features anywhere near as complex as XBL2? |
| 22:18 | <Philip`> | (particularly in terms of interactions with existing code, which presumably makes things much harder) |
| 22:18 | <othermaciej> | SVG is way more complex than XBL2 |
| 22:21 | <hsivonen> | http://www.w3.org/MarkUp/Forms/wiki/January-February_2008_Newsletter#head-f76a3e869082c8af2d5c659357614b51f5756ca9 |
| 22:22 | <Philip`> | Browser developers having been working on SVG for several years, and they're still all a bit rubbish at it |
| 22:23 | <roc> | othermaciej: that's arguable |
| 22:23 | <roc> | XBL2 is kind of invasive, it requires changes to a lot of core stuff |
| 22:23 | <annevk> | hsivonen, wow |
| 22:24 | <roc> | SVG is more modular, especially if you don't care about foreignObject really working |
| 22:24 | <Philip`> | (I gave up the last time I tried to do some dynamic SVG stuff, and switched to OpenGL, because things kept crashing or failing to redraw or silently failing in Opera and Firefox) |
| 22:24 | <othermaciej> | SVG wins by far on sheer size though |
| 22:24 | <annevk> | hsivonen, guess they redefined the meaning of "Forms TF" |
| 22:24 | <othermaciej> | and it does have some interaction complexities |
| 22:24 | <annevk> | SVG has <svg:use> |
| 22:24 | <annevk> | that's some simple XBL right there |
| 22:24 | <annevk> | in SVG :) |
| 22:24 | <roc> | I can tell you which of XBL1 and SVG causes *us* more maintenance headaches :-) |
| 22:24 | <othermaciej> | annevk: looks like clever shifting of what "we" refers to |
| 22:25 | <othermaciej> | <svg:use> is a giant crock of shit |
| 22:25 | <roc> | oh yes |
| 22:25 | <othermaciej> | it's so poorly designed compared to XBL, it's not even funny |
| 22:26 | Philip` | wonders what's wrong with it |
| 22:26 | <roc> | that's because we learned from all the mistakes in XBL1 |
| 22:26 | <othermaciej> | use has multiple references to a single underlying element |
| 22:26 | <othermaciej> | and at the point of use, you get things that aren't real Elements at all |
| 22:27 | <roc> | and style is supposed to cascade through each use point, somehow |
| 22:27 | <annevk> | true, XBL has had the unique opportunity to be redesigned from "scratch" without the usual Web compat issues |
| 22:27 | <othermaciej> | if you <use> a plugin multiple times for instance |
| 22:27 | <othermaciej> | each instance has to draw and process mouse events but they must all be in sync |
| 22:27 | <othermaciej> | in theory |
| 22:28 | <annevk> | roc, XBL2 has that feature as well (allowing selectors through and things like that) |
| 22:28 | <othermaciej> | I doubt anyone will ever bother to make that really work |
| 22:28 | <roc> | annevk: at least you don't have to have a variable number of computed styles applying to *the same element* |
| 22:28 | <othermaciej> | annevk: XBL2 multiply instantiates the shadow elements, rather than having the fake SVGElementInstance thing |
| 22:29 | <roc> | we actually clone anonymous nodes for <use> |
| 22:29 | <annevk> | ah ok, though updates to the binding have to be reflected everywhere it's bound |
| 22:38 | <Philip`> | Would browser developers willingly implement something as complex as SVG, now that they have some experience of the complexity that's involved? |
| 22:39 | <annevk> | it would've been useful if someone was there back in the days to decrease the feature creep |
| 22:48 | <annevk> | Philip`, http://diveintomark.org/archives/2008/03/09/no-fury-like-dracon-scorned#comment-11453 :p |
| 22:49 | Philip` | will have to start charging people for his services as an ill-formedness consultant |
| 22:49 | <annevk> | what a lame excuse from the other guy btw |
| 22:50 | eseidel | wonders how many people hacked on xbl |
| 22:51 | <Philip`> | annevk: "the other guy" = Aristotle? |
| 22:52 | jgraham | isn't sure call people viewing your page an "effective feedback loop" works so well when you are also hoping to call them "customers" |
| 22:52 | <jgraham> | s/call/calling/ |
| 22:55 | <Philip`> | I never give feedback to a site's owner, I just complain about their site on IRC or blog comments and it's not my problem if they don't notice it |
| 22:56 | <Philip`> | I suppose it's half a feedback loop, but the other half is just some ragged dangling ends that might happen to meet up if they're lucky |
| 23:00 | <Philip`> | It would be more effective to have some output filter which checks well-formedness of every HTTP reponse and emails the site owner if there's a problem, and then the filter might as well try to correct the error in a generally sensible way, and then you might as well not generate proper XML at all since you can rely on the filter to clean up after you |
| 23:09 | <Philip`> | and then you might as well find someone else to write the error-correcting part of that filter for you, preferably someone with lots of developers and QA people, like a web browser developer, and then you can call the filter "an HTML parser" |
| 23:17 | <Hixie> | hsivonen, annevk: i have to assume that they have something else they call the forms task force |
| 23:19 | <roc> | eseidel_: hacked on the XBL2 spec or hacked on the Gecko XBL1 implementation? |
| 23:22 | <eseidelDesk> | roc: gecko |
| 23:23 | <eseidelDesk> | roc: gecko xbl |
| 23:23 | <eseidelDesk> | roc: my point being that few have hacked on gecko svg |
| 23:23 | <roc> | few have hacked on XBL too really |
| 23:23 | <eseidelDesk> | roc: hence why I asked |
| 23:23 | <eseidelDesk> | since I don't actually know :) |
| 23:23 | <roc> | hyatt created it, then mainly sicking and bz |
| 23:23 | <eseidelDesk> | k |
| 23:25 | <roc> | SVG was probably broader, started by Alex Fritze and then tor, jwatt, longsonr, and me have all done significant work |