| 00:27 | <Philip`> | hsivonen: http://validator.nu/?doc=http://philip.html5.org/demos/html/ruby/wild-examples.html - why is it trying to parse the contents of the <xmp> elements? |
| 00:35 | <roc> | the next IE beta is apparently next quarter |
| 00:36 | <Philip`> | i.e. between a few days and three months away? |
| 00:36 | Philip` | wonders when quarters start |
| 00:36 | <Philip`> | Oh, probably a month plus a few days |
| 00:37 | <Philip`> | Silly 1-based month numbering system |
| 00:37 | <roc> | that's just how we do things on our planet |
| 02:31 | Dashiva | is tempted to bang head against wall |
| 02:32 | <Dashiva> | Maybe the alt discussion has as goal to induce self-inflicted injuries on watchers, to eliminate them from further participation |
| 05:06 | <Hixie> | annevk: i don't really mind you mentioning it on public-html |
| 05:07 | <Hixie> | it's just that every time anyone brings anything up on the public-html list, people always start asking things for the sake of purity instead of practical reasons |
| 05:07 | <Hixie> | e.g. the numerous questions about why <rb> isn't in the spec |
| 05:07 | <Hixie> | clearly none of the people who asked that have ever used ruby before |
| 06:08 | <MikeSmith> | Hixie: hey, it worked: "Discuss the difference between rel='' on <link> and <a>. (whatwg r1706)" |
| 06:08 | <MikeSmith> | thanks |
| 06:08 | <Hixie> | nice |
| 08:05 | <hsivonen> | Philip`: I have no idea why the contents of <xmp> are parsed. Do we have tree builder test coverage for xmp? |
| 08:40 | <annevk> | Hixie, seems your collegue did the same thing on the WHATWG list so it doesn't really matter :) |
| 08:40 | <annevk> | (and cc'ed the public-html list in the process...) |
| 08:45 | hsivonen | wonders how high-troughput apps deal with the eventual performance of server HotSpot |
| 08:49 | hsivonen | sees a 2D barcode on the desktop version of Nico Nico |
| 08:50 | <hsivonen> | Philip`: it seems that the xmp problem was fixed as a side effect of the major parser refactoring: http://validator.nu/?doc=http://philip.html5.org/demos/html/ruby/wild-examples.html |
| 09:03 | <annevk> | hsivonen, your parenthetical doesn't make sense (last e-mail) |
| 09:05 | <hsivonen> | annevk: hmm. looks like my turned itself off in the middle of the sentence :-( |
| 09:05 | <hsivonen> | see it happened again. whoa |
| 09:06 | <annevk> | It doesn't make much sense for "data-" to be conforming. You can't access it using dataset... |
| 09:06 | <annevk> | hmm, time for a new one? |
| 09:20 | <Philip`> | Hixie: Why isn't <rb> in the spec, for the sake of making existing pages that use ruby correctly with <rb> be conforming? :-) |
| 09:21 | <annevk> | Has anyone written a library for .dataset yet? |
| 09:21 | <annevk> | People are struggling with it: http://forum.mootools.net/viewtopic.php?pid=49569 |
| 09:22 | <annevk> | Other people hacking to add support for data-: http://creativepony.com/journal/2008/05/27/hack-markaby-to-let-you-use-a-html5-data-attribute/ |
| 09:23 | <Philip`> | hsivonen: Okay, thanks |
| 09:24 | <annevk> | Philip`, <rb> isn't in the spec because there's no need for it |
| 09:25 | <Philip`> | Hixie: s/affilated/affiliated/ in recent checkin |
| 09:27 | <Philip`> | annevk: It's needed for avoiding gratuitously making existing content non-conforming |
| 09:34 | <annevk> | Philip`, most content I've seen so far isn't conforming anyway :) |
| 09:40 | <Philip`> | annevk: Many are pretty close to being (machine-checkably) conforming, so it seems unhelpful to add unnecessary differences |
| 09:41 | <hsivonen> | would dropping <rb> and </rb> in the parser be bad? |
| 09:42 | <annevk> | yes, rb { color:red } and such |
| 09:42 | <Philip`> | That would be confusing, because of <ruby><rb style="font-family: whatever">stuff</rb>...</ruby> |
| 09:42 | <Philip`> | though it's effectively what IE does already |
| 09:42 | <Philip`> | (so people have to use <span> instead) |
| 09:42 | <annevk> | ruby::base { } |
| 09:42 | <annevk> | ruby::ruby-base { } |
| 09:43 | <hsivonen> | annevk: did you make those up just now or are they from the CSS draft? |
| 09:43 | <annevk> | made them up |
| 09:43 | <Philip`> | That seems unpleasantly complex |
| 09:44 | <Philip`> | It took me years to even work out you could have selectors like "parent child", and :: is much more confusing :-p |
| 09:44 | <annevk> | the alternative is <rb>, which makes the markup more complex |
| 10:48 | <Hixie> | Philip`: <rb> is non-conforming to help authors who may get trapped by trying to style <rb> without success |
| 10:50 | <annevk> | so dataset[""] = x doesn't work either? |
| 10:50 | <Philip`> | Hixie: By that logic you should make all of ruby non-conforming, to help authors who may get trapped by trying to use it when a reasonably significant number of users use browsers which don't yet support it |
| 10:51 | <Hixie> | annevk: per spec, it should "work" |
| 10:52 | <Hixie> | Philip`: yeah i didn't buy it either. the real reason it's not conforming is because i don't want to add a useless element to the spec, and ruby right now is in very early stages, so i'm willing to sacrifice a few authors today for the good of authors in the coming decades. |
| 10:53 | <annevk> | it does make implementing ruby slightly more complex |
| 10:53 | <Philip`> | Is there a known reason why the original ruby spec has <rb>? |
| 10:54 | <annevk> | apparently it's feasible to implement ruby using the inline table model from CSS and some to set a row as a baseline (Opera has a feature for that) |
| 10:54 | <Philip`> | annevk: Do you mean the presence of <rb> makes it more complex? |
| 10:54 | <annevk> | no, the absence |
| 10:55 | <Philip`> | annevk: You'd have to cope with people who don't use <rb> anyway, so its conformingness seems irrelevant |
| 10:56 | <annevk> | unless rendering is defined without requiring pseudo elements :) |
| 10:56 | <annevk> | but yeah |
| 11:09 | <Philip`> | annevk: Existing content omits <rb>, so, for compatibility, rendering would have to be defined without requiring explicit <rb> elements |
| 11:10 | Hixie | tries to support anne in the xhr debacle |
| 11:10 | <Philip`> | (I suppose they could be implied, like tbody etc, but that seems kind of nasty) |
| 11:11 | <annevk> | Hixie, I think that's going ok now. I just need to get a few outstanding things resolved and then I simply ask for WG resolutions on each because I can't work it out with Julian and then we can go to CR (or another LC, not sure) |
| 11:12 | <Hixie> | WG resolutions? |
| 11:12 | <annevk> | Philip`, seems the main problem is the rb/rt pairing in the current draft |
| 11:13 | <Hixie> | why can't you just make a decision and move on? |
| 11:13 | <Hixie> | you'll always have some peopel unhappy |
| 11:13 | <Hixie> | html5 has all kinds of things julian disagrees with :-) |
| 11:13 | <annevk> | Hixie, yes, but in order to suggest that he can raise a formal objection, i need support of the WG or something |
| 11:13 | <annevk> | Hixie, that seems like the simplest solution to me |
| 11:14 | Hixie | doesn't plan on trying to get "support of the WG" before telling people to raise formal objections :-) |
| 11:14 | Hixie | doesn't really care about formal objections at all in fact :-) |
| 11:15 | <annevk> | Hixie, ah, I was wondering whether to reply to that e-mail from sicking asking for more info |
| 11:17 | <Hixie> | seriously though, if you've considered all the input and come to a conclusion, and he just disagrees with your conclusion but agrees that there is no information you aren't aware of, then asking the wg isn't going to do any good |
| 11:17 | <Hixie> | groups of people make random decisions |
| 11:18 | <Hixie> | the more people the more random |
| 11:18 | <Hixie> | random decisions aren't what's best for hte web |
| 11:18 | <annevk> | I had this debacle once in the CSS WG, I don't want it again |
| 11:19 | <annevk> | I don't like it when chairs of WGs go all mad on me |
| 11:19 | <Hixie> | eh you'll get used to it |
| 11:20 | <Hixie> | i guess i'm in the lucky position of being able to just pull out of the w3c altogether if it gets too crazy |
| 11:20 | <Hixie> | we really need a patent policy setup for whatwg |
| 11:20 | <Hixie> | that way we wouldn't need the w3c at all |
| 11:23 | <gsnedders> | Silly W3C. |
| 11:23 | <gsnedders> | "The only person who counts is the editor. If you disagree with the editor, go tell your mummy on him." |
| 11:25 | <Hixie> | if enough people disagree with the editor, the editor gets replaced and is out of a job |
| 11:25 | <Hixie> | but you'll always find people who disagree |
| 11:25 | <Hixie> | anyway. bed time. nn. |
| 11:25 | <gsnedders> | Nah. Everyone agrees with me. |
| 11:25 | <gsnedders> | night, Hixie |
| 11:59 | Philip` | really doesn't like when online purchase systems show a payment page with an iframe containing (what claims to be) a page from his bank asking for his number and password for confirmation |
| 12:00 | <Philip`> | since I can't tell where the page is actually from, and even when I can tell I just find it's www.securesite.co.uk and can't tell how legitimate that is |
| 12:01 | <Philip`> | (I'm not even sure if there's an easy way to tell that it's HTTPS) |
| 12:03 | <annevk> | the banking system in the Netherlands uses redirects for that |
| 12:04 | <Philip`> | Oh, actually, securesite.co.uk is something totally unrelated - the legitimate one is securesuite.co.uk |
| 12:06 | <Philip`> | (http://ambrand.com/2006/09/06/is-securesuitecouk-a-phishing-scam/ etc) |
| 12:50 | <gsnedders> | Philip`, jgraham: Are either of you involved in the uni open days on 3rd/4th July? |
| 12:51 | <Philip`> | gsnedders: I'm not planning to be |
| 12:53 | <Philip`> | (though I'm also not planning not to be) |
| 12:54 | <gsnedders> | Philip`: :) |
| 12:54 | gsnedders | will be there |
| 12:54 | <Philip`> | (Depends on whether someone from our group cares enough to do a demo or something) |
| 12:55 | <gsnedders> | I need to decide whether to have a look at the Nat.Sci. stuff or not |
| 12:56 | <Philip`> | (Oh, I guess we don't care enough, since the email said to reply by 2nd May if interested in demoing) |
| 12:57 | <Philip`> | Any particular NatSci stuff? |
| 12:59 | <gsnedders> | Philip`: Well, Physics may be a good guess seeming that's what I've talked about |
| 13:00 | <gsnedders> | Philip`: and seeming I haven't done any other pure sciences in years (I regret not doing Chem., but it's too late now)… |
| 13:03 | <Philip`> | gsnedders: Ah, sounds reasonable |
| 13:04 | <gsnedders> | The other two prospectuses (York and Edi) arrived today, FWIW |
| 13:04 | Philip` | still hasn't worked out how to extract useful information from open days |
| 13:04 | <Philip`> | Both today? Sounds like a conspiracy |
| 13:24 | <gsnedders> | Philip`: Well, it still stands true that Cam's propaganda department is better |
| 13:25 | <Philip`> | Hmm, I haven't physically used my computer at the CL for so long that I've forgotten the login password :-/ |
| 13:37 | <gsnedders> | Philip`: "password"? |
| 13:39 | <Philip`> | gsnedders: http://en.wikipedia.org/wiki/Password |
| 13:41 | <gsnedders> | Philip`: Why do I think you deliberately misinterpreted what I meant? |
| 13:41 | <Philip`> | (Turns out that actually I did remember the password, but the magic networking stuff was broken so it didn't recognise it until I rebooted (and it takes forever for the computer to reboot because it's got lots of magic networking stuff to set up)) |
| 13:41 | <Philip`> | gsnedders: I didn't know how to correctly interpret what you wrote |
| 13:42 | <Philip`> | (and still don't) |
| 13:42 | <gsnedders> | Philip`: The string password is your password. |
| 13:42 | <Philip`> | Oh |
| 13:42 | <Philip`> | It isn't :-p |
| 13:47 | Philip` | now has Synergy working, so he can use his laptop's mouse and keyboard to control his desktop computer (which has 4x as much screen area, which is sometimes handy) |
| 13:47 | <gsnedders> | Philip`: See, this is why I have a 17" laptop :P |
| 13:49 | <Philip`> | I've got a 15" laptop, which usually isn't too small and isn't too hard to carry around :-) |
| 13:49 | <Philip`> | but it's not so good when I want a dozen virtual machine shell windows at once |
| 13:51 | virtuelv | has a 12.1" laptop |
| 13:51 | <virtuelv> | (And an external 20" widescreen monitor) |
| 13:53 | <gsnedders> | I just don't carry mine around much :) |
| 13:53 | <gsnedders> | And I have a nice rucksack to carry it in now :) |
| 14:26 | <gsnedders> | I'm about to make myself look really stupid. Time for me to try and write C. |
| 14:28 | <gsnedders> | I want to see how much I can speed html5lib up by just rewriting inputstream.charsUntil in C |
| 14:36 | <annevk> | algirht, now I need to move all XHR1 changes to XHR2 |
| 16:03 | <gsnedders> | I'm failing badly at getting charsUntil any quicker |
| 16:07 | <Philip`> | Which part of charsUntil is slow? |
| 16:08 | <Philip`> | like, is it doing lots of complex calculation itself, or is it just spending its time pulling characters out of Python data structures? |
| 16:09 | <gsnedders> | Philip`: As far as I can see, the former |
| 16:09 | <gsnedders> | well, it's hard to say :P |
| 16:11 | <Philip`> | Try removing all the functionality from the function and see how that affects performance :-) |
| 16:12 | <Philip`> | and if it's still just as slow, then it must be the interface with surrounding code that's slow, in which case I guess you can't do much to optimise that function by itself |
| 16:13 | <gsnedders> | Philip`: There isn't really much functionality to remove, without completely affecting what the function has to do. Removing some of the stuff would make it far quicker as it was operating on less data. :P |
| 16:14 | <annevk> | just rewriting it in C would help |
| 16:15 | <Philip`> | You could remove all the stuff that implements the "until" thing, and run it with test data that doesn't depend on the "until" thing working, and see how much effect that has |
| 16:15 | <annevk> | and then we need it to ship with python by default |
| 16:31 | <Philip`> | Seems like charsUntil can be made ~30% faster just by using a list instead of a deque |
| 16:31 | <Philip`> | (and storing the string reversed, so all the operations happen on the tail of the list) |
| 16:32 | <Philip`> | (except for readChunk but that can be made to only be called when the list is nearly empty, I think) |
| 16:37 | <Philip`> | (Also it could maybe be made much faster by storing a list of strings, instead of a list of characters) |
| 16:37 | Philip` | might look at that later tonight |
| 16:38 | <gsnedders> | Philip`: How would that allow you to make it quicker, thoguh |
| 16:38 | <gsnedders> | *though |
| 16:38 | <Lachy> | oh crap! I just lost a whole heap of slides cause Keynote died :-( |
| 16:38 | <Lachy> | I |
| 16:38 | <Lachy> | I'll have to remake them now :-( |
| 16:38 | <Philip`> | gsnedders: Which "that" do you mean? |
| 16:38 | <gsnedders> | Philip`: str not char |
| 16:38 | <Philip`> | Lachy: Doesn't all office software have autosave? |
| 16:38 | <Philip`> | gsnedders: It'd let you run regexps |
| 16:39 | <Philip`> | and depending on how Python stores lists, it might be hugely more memory-efficient (hence cache-efficient and everything) |
| 16:39 | <gsnedders> | Philip`: http://code.google.com/p/html5lib/issues/detail?id=69 |
| 16:40 | Philip` | wishes Opera didn't insist on opening patch files with "less" |
| 16:40 | <gsnedders> | Philip`: What if it opened it with "more"? |
| 16:40 | <hsivonen> | Lachy: no Time Machine? |
| 16:40 | <Philip`> | gsnedders: It'd let you run regexps without wasting loads of time calling 'join' |
| 16:41 | <Philip`> | (and without wasting memory etc) |
| 16:41 | <takkaria> | are strings immutable in python? |
| 16:41 | <Dashiva> | Philip`: Maybe you shouldn't set it to use less then :) |
| 16:42 | <Philip`> | Dashiva: I never did, since that'd be stupid :-) |
| 16:42 | <Philip`> | takkaria: Yes |
| 16:42 | <Lachy> | hsivonen, no. The weird thing is I thought I had saved it regularly, but I lost everything since last night |
| 16:42 | <gsnedders> | Philip`: It probably means changing a heckuva lot though |
| 16:42 | <Lachy> | Philip`, obviously keynote doesn't, though I wish it would |
| 16:43 | <Philip`> | gsnedders: I can't see any references other than a dozen mentions of self.queue in inputstream.py, and it should be feasible to optimise things somewhat without changing its interface |
| 16:46 | <Lachy> | at least this time I can remake it faster since I don't have to make decisions again, just remember what I had done, and I still have all the images |
| 17:09 | <annevk> | gsnedders, the real solution here is a port of hsivonen's parser to C++ and make a Python binding for that and get it shipped with Python 2.x |
| 17:10 | <annevk> | gsnedders, though maybe that's too complex for now |
| 17:10 | <gsnedders> | annevk: Yeah, sure. That gets a quick parser. But how about when you want to iterate over every element in HTML 5? libxml (and lxml therefore too) is the only thing reasonably quick currently |
| 17:25 | <annevk> | http://www.w3.org/Submission/2008/01/ |
| 17:31 | <gsnedders> | HTTP is really odd. |
| 17:32 | <gsnedders> | I find it very interesting that IIS (or HTTP.sys, really) is so much stricter than Apache |
| 17:33 | <gsnedders> | Even at a basic level, Apache accepts "HTTP / 1.1" while IIS sends 400 |
| 17:34 | <gsnedders> | it is, arguably, allowed by RFC2616 |
| 17:34 | <gsnedders> | implied lws sux. |
| 17:39 | <Dashiva> | annevk: So they want to push a patented algorithm into web fonts, or? |
| 17:41 | <annevk> | beats me |
| 17:41 | <annevk> | with the recent advanced of Opera and Safari, and soon Mozilla, I think EOT is pretty much dead |
| 17:41 | <annevk> | advancements* |
| 17:42 | <takkaria> | except for people who care about IE |
| 17:46 | gsnedders | reports a bug on CFNetwork |
| 17:51 | <Lachy> | any suggestions for what I should put on the slide about <video>? We're probably going to show an actual demo video in a browser for this as well |
| 17:51 | Philip` | discovers that the bottleneck in his code is "for i in range(self.chunkOffset, len(self.chunk)):", so he rewrites it as an ugly "i = ...; while i < ...; ...; i += 1" and cuts 40% off his execution time |
| 17:52 | <Philip`> | (I would have thought Python wouldn't be so incredibly dumb as to fail to optimise that very common case) |
| 17:53 | <Dashiva> | Nobody uses indexes anymore, Philip`. It's all lists. ;) |
| 17:53 | <Philip`> | The whole point of this code is that lists are slow and I want to avoid them :-p |
| 17:53 | <Lachy> | Philip`, do you have any other cool <canvas> demos we could show off in our presentation? |
| 17:54 | <hober> | Philip`: xrange is your friend |
| 17:54 | <Lachy> | I might just use your game at canvex.lazyilluminati.com |
| 17:55 | <Philip`> | hober: Aha, thanks, that's not noticeably slower than the manual loop, which is nice :-) |
| 17:55 | <Philip`> | (though it'd still be nicer if Python did that optimisation automatically) |
| 17:56 | <Philip`> | Lachy: Canvex is about the only significant <canvas> thing I've done |
| 17:56 | <Lachy> | Philip`, ok. |
| 17:56 | <Philip`> | Lachy: There's http://www.p01.org/releases/DHTML_contests/files/20lines_twinkle/ which might be interesting |
| 17:56 | <Lachy> | Philip`, could you zip up all the files for me so I can easily download and use it offline? |
| 17:56 | <annevk> | Lachy, ask p01, he made that |
| 17:57 | <Philip`> | Lachy: Also http://glimr.rubyforge.org/cake/canvas.html with various stuff |
| 17:57 | <Philip`> | Lachy: What annevk said :-) |
| 17:58 | <Lachy> | annevk, yeah, I will. But was asking Philip` for his canvex game |
| 17:58 | <Philip`> | Lachy: Oh, I misinterpreted you in the same way |
| 18:00 | <Philip`> | Lachy: http://canvex.lazyilluminati.com/83/83.zip is an almost recent copy of hopefully all the necessary files |
| 18:00 | <Philip`> | (I think it's just missing the Safari logo in the outdoors bit, because the game didn't work in Safari back then) |
| 18:03 | <Lachy> | Philip`, it's reporting a well formedness error in ./textures/misc/ui_bottom.svgz |
| 18:03 | <Philip`> | Lachy: That's because Firefox is stupid, I think |
| 18:04 | <Philip`> | Either run it on a local web server, or modify play.xhtml to refer to .svg instead of .svgz |
| 18:04 | <Philip`> | ...and if the .zip doesn't include .svg files then gzip -d the .svgz ones |
| 18:04 | <Lachy> | what's the .svgz file? |
| 18:04 | <Philip`> | gzip-compressed SVG |
| 18:04 | <Lachy> | ok, I can run it from a webserver |
| 18:05 | <Philip`> | Might need to make sure the web server is configured to send the right content-type/encoding for .svgz - I'm not sure if they do it by default |
| 18:06 | <Philip`> | (The problem with using file:/// is that Firefox guesses the content-type but not -encoding for .svgz files, so it tries parsing the compressed data as XML) |
| 18:06 | <Lachy> | what's the apache directive for setting the content-encoding? |
| 18:08 | <Lachy> | Philip`, what's the purpose of the graph I get when I press Cmd+T, and why have you overridden my Open New Tab command? |
| 18:08 | <Philip`> | AddEncoding gzip .svgz |
| 18:08 | <Philip`> | Lachy: It shows the framerate, and I overrode your new tab command because I'm cruel and unthoughtful |
| 18:33 | <gsnedders> | ARGH! |
| 18:33 | gsnedders | goes mad at RFC2616 using a totally undefined form of BF |
| 18:33 | <takkaria> | someone stamp on your foot? |
| 18:33 | <gsnedders> | *BNF |
| 18:53 | <Lachy> | oh, damn. running Windows in VMWare on Mac, going to http://localhost/ doesn't see the server running on the Mac, it tries to find the server running on Windows. :-( |
| 18:55 | <hober> | ipconfig /all in a terminal, look at the IP for 'default gateway', and point the browser at that |
| 18:55 | <hober> | IIRC |
| 18:56 | <gsnedders> | Lachy: Change the network type |
| 18:57 | <Lachy> | thanks hober |
| 18:59 | <Lachy> | gsnedders, ? |
| 19:00 | <gsnedders> | Lachy: IIRC, if you go to Settings -> Network (on the VM) one of the options should make localhost the host |
| 19:00 | <Lachy> | hober, re http://krijnhoetmer.nl/irc-logs/whatwg/20080515#l-1206 - did you get a chance to follow up on that? |
| 19:02 | <hober> | not yet; work's been pretty crazy |
| 19:03 | <hober> | hoping to take a look at it this weekend |
| 19:04 | <Lachy> | gsnedders, none of the prefs in there seem to help |
| 19:10 | <Philip`> | Lachy: You could use your real machine rather than a VM |
| 19:16 | <Lachy> | Philip`, I need to use the windows build because we don't have a suitable mac build that supports video |
| 19:16 | <Philip`> | Oh, how annoying - I can easily make charsUntil ten times faster on my artifical test case, but it makes about zero different in practice :-( |
| 19:17 | <Philip`> | *difference |
| 19:17 | <Philip`> | Lachy: Ah, okay |
| 19:21 | <Philip`> | ...and when parsing the spec (with parse.py --no-html), inputStream.py is only about 30% of the total execution time anyway, so there's little room for global improvement anyway |
| 19:23 | Philip` | gives up instead |
| 19:26 | <gsnedders> | Philip`: That's why I was trying with the spec. That's not so artificial. |
| 19:26 | <Philip`> | Actually I'm cheating and testing with the first 1K or 10K lines of the spec, since I'm lazy |
| 19:27 | <gsnedders> | Philip`: parser.parse() with the lxml etree builder makes charsUntil 70% of the total time |
| 19:31 | <Windstoss> | Hi there! |
| 19:32 | <Windstoss> | Does Web Applications 1.0 either directly or indirectly refer to the Selectors API or XmlHttpRequest of the W3C Web API WG? |
| 19:33 | <Philip`> | gsnedders: When I run parse.py --profile --no-html --treebuilder=lxml on the first 10K lines of the spec, it spends cumtime=0.819 out of total 5.982 in charsUntil |
| 19:34 | <Philip`> | (and 1.507 in total in inputstream.py, vs 2.116 in tokenizer.py and 1.019 in html5parser.py) |
| 19:35 | <Windstoss> | This Colloquy sort of IRC Client is strange :( |
| 19:35 | <Philip`> | Windstoss: It does |
| 19:35 | <Philip`> | Windstoss: e.g. http://www.whatwg.org/specs/web-apps/current-work/multipage/section-origin.html has an [XHR] reference |
| 19:36 | <Philip`> | Windstoss: (Doesn't refer to the Selectors API as far as I can see, though) |
| 19:37 | <Windstoss> | Philip`: thanks! :) |
| 19:37 | <Windstoss> | Philip`: I'm writing a thesis about HTML 5 and was wondering, if the Selectors API is something seperate |
| 19:38 | <gsnedders> | Philip`: That's totally different to what I get from cProfile… |
| 19:38 | <Philip`> | Windstoss: I think the only relation is that similar people are working on both |
| 19:38 | <gsnedders> | Philip`: 5.214 out of 17.312 using python parse.py --no-html --profile --treebuilder=lxml ../../html5 |
| 19:39 | <gsnedders> | Philip`: (that's on an entire several month old .src.html copy of html5) |
| 19:39 | <Windstoss> | Philip`: you mean those pragmatic, we do whats necessary sorts of people? ;) |
| 19:41 | <Philip`> | gsnedders: Whoops, I was using my modified inputstream.py... With the original it's more like 1.542 out of 6.628 in charsUntil |
| 19:41 | <Philip`> | but that's still only ~25% |
| 19:41 | <Philip`> | Oh, which is not much different to what you saw |
| 19:41 | <Philip`> | (but is quite different to 70%) |
| 19:42 | <gsnedders> | I got 70% from cProfile |
| 19:43 | <Philip`> | Even cutting execution time by 70% wouldn't make html5lib significantly faster, compared to the orders of magnitude from writing the whole thing in Java/C/etc |
| 19:44 | Philip` | gets depressed and goes home |
| 19:44 | <gsnedders> | Sure, but some of us don't know Java/C :P |
| 19:44 | <Philip`> | That's why you start collecting vast amounts of money to encourage a Java/C programmer to write it for you |
| 19:45 | <gsnedders> | Philip`: Or I just learn C :) |
| 19:46 | <gsnedders> | HTTP is _so_ odd. |
| 20:11 | Dashiva | wonders if JF realizes he's arguing that valid HTML is useless |
| 20:26 | <takkaria> | I love facebook |
| 20:26 | <takkaria> | in terms of availability, it's getting to be worse than myspace |
| 21:24 | <Hixie> | is w3.org down for anyone else? |
| 21:24 | <billmason> | Yes. |
| 21:25 | <Hixie> | i swear the w3.org services are down more often than whatwg.org |
| 21:26 | <hober> | w3.org is working for me |
| 21:26 | <hober> | at least, browsing/reading mailing list archives |
| 21:27 | <Hixie> | those are on lists.w3.org |
| 21:27 | <Hixie> | different server |
| 21:27 | <Philip`> | www.w3.org (in France) doesn't work for me |
| 21:27 | <Philip`> | Oh wait, maybe it does |
| 21:27 | <Philip`> | except not over the web, and traceroute goes funny |
| 21:28 | <Hixie> | funny how? |
| 21:29 | <Philip`> | It just gets to inria-nice.cssi.renater.fr and then starts printing *s |
| 21:29 | <Philip`> | which is not particularly funny |
| 21:30 | <Hixie> | ah |
| 21:30 | <Hixie> | that probably just means the next host is down |
| 21:30 | <Hixie> | (or blocking ICMP) |
| 21:30 | <smedero> | it is back up |
| 21:30 | <Hixie> | so it is |
| 21:30 | <Philip`> | ping worked, so it was probably just filtered |
| 21:30 | Lachy | is glad the spec is mirrored! :-) |
| 21:30 | <Hixie> | Lachy: sadly, w3.org's mailing list search is not :-( |
| 21:31 | <Lachy> | it's ok. My email on dreamhost is down again as well |
| 21:31 | <Hixie> | heh |
| 21:31 | <Lachy> | so I hadn't noticed |
| 21:32 | <Lachy> | any idea what photo I could use to represent video on the <video> slide? |
| 21:32 | <Hixie> | a dog on a skateboard |
| 21:32 | <Lachy> | I'm looking for a painting canvas, possibly sitting on an easel for use on the <canvas> slide |
| 21:32 | <takkaria> | cinefilm rolls, or something of that nature |
| 21:33 | <Philip`> | Lachy: A screenshot from http://www.youtube.com/watch?v=oHg5SJYRHA0 |
| 21:33 | <Lachy> | Hixie, any particular reason, or did you just pick something random? |
| 21:33 | <Lachy> | LOL! |
| 21:33 | <Hixie> | that's like the canonical youtube video |
| 21:33 | <Hixie> | (the dog on the skateboard) |
| 21:33 | <Dashiva> | I thought the canonical youtube video was a cat doing nothing |
| 21:33 | <Hixie> | a screenshot from the rick roll video would be pretty funny to |
| 21:33 | takkaria | curses Philip` |
| 21:33 | <Lachy> | yeah, a photo of Rick Astley would be awesome |
| 21:34 | <Hixie> | o |
| 21:34 | <Dashiva> | So will the <video> part be a rickroll? |
| 21:35 | <Lachy> | I should get a copy of that video for the presentation |
| 21:35 | <Lachy> | though, I was intending to use the video of me skiing over a bus |
| 21:36 | <Lachy> | http://itsnotpossible.typepad.com/trashfan/astley1.jpg :-) |
| 21:36 | <Philip`> | Someone managed to make that video play during a lecture (on 'Introduction to Security') here recently, by sneaking in at night and booting the lecture theatre's computer into Linux to copy a script into Windows' Startup folder to play the video when the lecturer was logged in, which was a useful demonstration of the need for more careful security :-) |
| 21:38 | <Lachy> | LOL |
| 21:39 | <smedero> | when I'm looking to break up a text heavy presentation involving HTML, I often nab a photo from the visual HTML jokes flickr pool: http://www.flickr.com/groups/htmljokes/pool/page5/ |
| 21:41 | <Lachy> | smedero, I've already got 2 LOL cats, a page of Corporate Ipsum, a Rick Astley photo (possibly a video too) |
| 21:41 | <smedero> | sounds like you've got it covered. :D |
| 21:43 | <Lachy> | I can't find a decent picture of a canvas on an easel with an appropriate licence. :-( |
| 21:43 | <Lachy> | maybe I should just go for some famous painting |
| 21:44 | <gsnedders_mibbit> | And me! Don't forget me! |
| 21:44 | <gsnedders_mibbit> | (of what's in the presentation) |
| 21:44 | <Lachy> | gsnedders_mibbit, you're still in there |
| 21:45 | <Lachy> | though I still have to fit the drunk MikeSmith photo in somewhere too |
| 21:47 | <Dashiva> | I wonder if there's a lorem ipsum equivalent for IDE designers, that uses lorem ipsum generator code |
| 21:48 | <hdh> | they can just ROT13 the IDE's source |
| 21:51 | <Lachy> | Dashiva, where would an IDE designer user lorem ipsum? |
| 21:54 | <Lachy> | I now have a painting of FSM on the canvas slide |
| 21:55 | <Lachy> | the one Michelangelo painted on the Sistine Chapel :-) |
| 21:55 | <Philip`> | Lachy: I hope you're bearing in mind that HTML is Serious Business |
| 21:56 | <Lachy> | Philip`, yeah, just gotta keep people interested so they'll listen to the serious stuff |
| 22:25 | <takkaria> | hmm, RB accuses Hixie of ignoring some browser implementers |
| 22:31 | <Philip`> | It seems true that there hasn't been much input from e.g. Lynx developers, or those people who make mobile browsers that don't work as well as Opera/WebKit, or AT developers who build on top of existing browsers |
| 22:31 | <takkaria> | tis hardly something Hixie can help though |
| 22:32 | <takkaria> | I dunno, maybe I should just killfile RB and let him be someone else's problem |
| 22:32 | <Lachy> | I like how, when we attempted to address RB's concerns, he raised the bar and bitched some more http://lists.w3.org/Archives/Public/public-html/2008May/0622.html |
| 22:33 | <Lachy> | like, complaining it was added to a later draft, instead of the revision the survey was about, kind of misses the point of asking for someting to be added |
| 22:34 | <takkaria> | the comments about consensus-building are interesting |
| 22:34 | <takkaria> | (in the WG questionnaire) |
| 22:34 | <Philip`> | Perhaps his concerns could be addressed by saying anything that isn't already mentioned in the spec (including the various placeholder notes) isn't going to be added in the future, so he doesn't have to worry about major features not being present yet |
| 22:36 | <takkaria> | http://html4all.org/mailman/archives/list_html4all.org/2008-May/000873.html terrifies me |
| 22:37 | <Lachy> | maybe I should start complaining more loudly. There are still features that I've asked for in the past which still haven't been added. |
| 22:37 | <Philip`> | takkaria: In terms of the web, that seems to be the wrong place to concentrate effort - approximately nobody is going to use such complex markup, and most of those people will get it wrong, so there's a much better return on investment by developing decent heuristics for pronouncing un-marked-up text |
| 22:39 | <takkaria> | Philip`: I knew that already, but maybe now you'll be quoted on the private html4all list |
| 22:39 | <takkaria> | RB is, I think, worse than Dmitry for these kinds of proposals now |
| 22:40 | <Philip`> | takkaria: That private list is only used for administrative details so I'm sure they wouldn't conduct the kind of private conversations there that they used to conduct on list⊙ho |
| 22:40 | <takkaria> | ah |
| 22:40 | <takkaria> | http://html4all.org/mailman/archives/list_html4all.org/2008-May/000856.html and the messages immidiately after it are similarly astronauty proposals |
| 22:40 | <takkaria> | brb, anyway, popping to tesco |
| 22:41 | <Lachy> | what is tesco? |
| 22:41 | <Philip`> | There's nothing necessarily bad about having markup for pronunciations of text, in certain contexts; it's just not appropriate for the web, which is what HTML5 is focussing on |
| 22:41 | <Philip`> | Lachy: http://en.wikipedia.org/wiki/Tesco |
| 22:41 | <Philip`> | (Big supermarket) |
| 22:42 | <Lachy> | wow, didn't realise they were so big |
| 22:42 | <Lachy> | surprising that I hadn't heard of them before |
| 22:43 | <Philip`> | "over £1 in every £7 [...] of UK retail sales is spent at Tesco" |
| 22:46 | <Lachy> | how can I represent offline web apps with a diagram? |
| 22:47 | <Philip`> | Gmail logo + unplugged network cable? |
| 22:47 | <Lachy> | hmm. A photo of an unplugged network cable would be nice |
| 22:47 | <Lachy> | that's something I can easily take a photo of myself, so even better |
| 22:55 | <annevk> | Lachy, you want http://www.flickr.com/photos/f8dy/347023948/ |
| 22:59 | <Lachy> | ah, right. I think I saw a better one of those last week |
| 23:01 | <hober> | takkaria: my linguist wife said, re: the link that terrifies you, "uugggggggh. this guy needs to go back to Lojban land." |
| 23:01 | <Lachy> | wtf? I don't get this one. http://www.flickr.com/photos/jotape_es/384293705/in/pool-htmljokes |
| 23:02 | <takkaria> | hober: heh |
| 23:03 | <takkaria> | Lachy: http://en.wikipedia.org/wiki/Font_(disambiguation) |
| 23:03 | <annevk> | a whole new WG just for geolocation? |
| 23:04 | <Lachy> | takkaria, oh, so it's old word for fountain |
| 23:04 | <Lachy> | which still doesn't make that much sense, since I would call that a bubbler |
| 23:04 | <annevk> | imagine if we needed separate WGs for HTML syntax, SQL storage, persistent storage, offline/online events, offline application cache, sandboxing, "origin", etc. |
| 23:04 | Philip` | has never heard the term "bubbler" for fountains before |
| 23:05 | <Lachy> | http://en.wikipedia.org/wiki/Bubbler |
| 23:05 | <Philip`> | (Actually, I'm not sure I've heard the term "bubbler" ever used seriously for anything) |
| 23:05 | <Philip`> | Crazy foreigners :-p |
| 23:06 | <hober> | Philip`: heh. /me speaks en_US-x-boston |
| 23:14 | <Philip`> | Is it intentional that I get 13 errors when running the html5lib tests? |
| 23:15 | <Dashiva> | Hixie isn't shiny enough, he doesn't reflect the WG |
| 23:17 | <takkaria> | let's replace him with an inanimate piece of shiny silver |
| 23:18 | <Hixie> | i can tell i'm going to have fun when i read today's htmlwg mail |
| 23:18 | <Dashiva> | No, just the one by RB, Hixie :) |
| 23:18 | <Dashiva> | Then again, that one mail is plenty |
| 23:19 | <Dashiva> | Well, there's also the different accessibility guys contradicting each other left and right, but that's nothing new |
| 23:20 | <takkaria> | and MMM's latest post is amusing |
| 23:21 | <Dashiva> | "Do you expect UAs to implement acutal useful features? This is an outrage!" |
| 23:21 | <Dashiva> | (Not part of today's mail) |
| 23:21 | <annevk> | Hixie, do you know if sicking is just replying to the AC questions one by one? |
| 23:22 | <annevk> | Hixie, he had objections to all three, but he's only starting a new thread on one |
| 23:22 | <Hixie> | no idea |
| 23:22 | <annevk> | sigh |
| 23:22 | <Hixie> | i propose you ignore feedback that has had no reasoning |
| 23:22 | <Hixie> | and say so |
| 23:23 | <Philip`> | (Hmm, my html5lib now parses the spec (and discards the output) 15% faster than before, and doesn't fail any tests that it didn't fail yesterday) |
| 23:23 | <annevk> | sure, i just don't want to get bits and piece feedback for the next few weeks |
| 23:24 | <annevk> | Mozilla has surprisingly similar stalling tactics to Microsoft, though I guess it's not intentional |
| 23:24 | <Dashiva> | Maybe MS should file a patent |
| 23:24 | <annevk> | Actually, I know it's not intentional (at least until now), sorry about that |
| 23:25 | <Philip`> | Nobody apologises when accusing Microsoft of anything :-p |
| 23:27 | <Dashiva> | Philip`: Well, there's enough reasonable accusations to make that there's seldom a need to make unreasonable ones ;) |
| 23:28 | <annevk> | It's kind of frustrating though. sicking says he'll give feedback the same day. Then waits three days and provides only a third of what he promised. |
| 23:29 | <gavin_> | have you asked him about it? |
| 23:30 | <annevk> | Hixie did |
| 23:48 | <Lachy> | What picture should I use to represent software like html5lib and validator.nu? |
| 23:51 | <Lachy> | are there any other software projects I should mention besides those two? |
| 23:51 | <gsnedders> | Don't _ever_ forget to install Boot Camp 2.1 before upgrading to XP SP3 on Apple hardware. |
| 23:51 | gsnedders | has just spent hours digging himself out of that hole |
| 23:52 | <Dashiva> | Lachy: The test case projects? |
| 23:53 | <Lachy> | Dashiva, stuff that's relevant to develpers |
| 23:53 | <Lachy> | test cases are only relevant to implementers |
| 23:54 | <gsnedders> | Lachy: I'm not relevant to developers :P |
| 23:54 | <Dashiva> | I suppose most of the other language implementations are personal projects |
| 23:54 | <Lachy> | gsnedders, your photo is the main attraction ;-) |
| 23:55 | <gsnedders> | Lachy: :D |
| 23:55 | <takkaria> | Lachy: if you want html5lib as a library, then maybe a library... |
| 23:55 | <takkaria> | if you mean html5lib as a suite of tools, I dunno |
| 23:55 | <Lachy> | I'd prefer something that represents it as a development tool |
| 23:56 | <Dashiva> | A toolbox is a bit cliche :) |
| 23:56 | <takkaria> | bricks and mortar are a traditional kind of development tool icon |
| 23:56 | <takkaria> | at least where I come from |
| 23:56 | <Hixie> | lachy: http://www.buytikitorches.com/images/funnel-copper_1.jpg |
| 23:56 | <Hixie> | with random garbage pouring in the top |
| 23:56 | <Hixie> | and tiny beautiful trees coming out of the bottom |
| 23:57 | <Lachy> | Hixie, assume I don't have the time or skills to make such modifications |
| 23:57 | <Philip`> | http://www.accessibility.nl/files/images/tag-soup2-035.jpg |
| 23:57 | <Dashiva> | accessibility isn't a tag! |
| 23:58 | <Dashiva> | "Waiter, there're buzzwords in my tag soup" |
| 23:58 | <Lachy> | hah :-D |
| 23:58 | <Philip`> | When's Google going to make image search that actually works, and shows me tag soup instead of American 'football' players with tins of CHUNKY Savory Pot Roast and suchlike? |
| 23:59 | <Hixie> | Dashiva: i'm sure there are pages with <accessibility> tags! |
| 23:59 | <Dashiva> | <accessibility><html>...</html></accessibility> solves everything |
| 23:59 | <takkaria> | no, making alt mandatory solves everything |
