| 00:05 | <zewt> | "not supported" depending on state sort of seems odd |
| 00:06 | <Hixie> | InvalidStateError then? |
| 00:06 | <zewt> | InvalidStateE...yes |
| 00:06 | <Hixie> | k |
| 00:25 | <Hixie> | ok, <dialog> is taking shape |
| 00:26 | <Hixie> | probably be done tomorrow, if no hidden surprises come up |
| 00:27 | <tantek> | hidden surprise dialogs are no good |
| 00:27 | <Hixie> | heh |
| 00:27 | <Hixie> | oh, wait, i forgot about hte magic form stuff |
| 00:27 | <Hixie> | that'll take longer |
| 00:29 | <Hixie> | <form method=dialog> <input type=submit> </form> |
| 00:29 | <Hixie> | where the submit button closes the dialog and sets dialog.returnValue to submit button's value |
| 00:30 | <Hixie> | with all the form validation stuff happening as well of course |
| 00:30 | <Hixie> | hmm |
| 00:31 | <Hixie> | or i guess we could just have them use <form action="javascript:void(0)" onsubmit="parentElement.close()"> |
| 00:31 | <Hixie> | but that seems lame |
| 00:32 | <Hixie> | then again, maybe instead of method=dialog, we should have method=none so it would work outside dialogs too |
| 00:32 | <Hixie> | and just do the closing-dialog behaviour be a magic thing if you happen to be inside a form? |
| 00:32 | <Hixie> | ok i'll think about it. if anyone has any ideas, paste them here or on the wiki. |
| 00:32 | <Hixie> | bbl. |
| 01:32 | <tantek> | Hixie - regarding having a form submit button close the dialog etc., could we do the same for modeless pop-up-window "dialogs"? That is, have some way for a form in a child window to close the window upon successful submission? |
| 01:33 | <tantek> | The specific use-cases I'm thinking of here are quite similar to the dialog use-cases, e.g. Plancast.com pops-up a twitter sign-in window, and many tweet buttons/links also pop-up a window for the user to complete their tweet and submit. Would be great if that submit could somehow close the child window. |
| 01:34 | <tantek> | (without requiring JS to be one for it to work, note Twitter's tweet actions themselves work without JS but the pop-up windows still require the user to manually close them when done) |
| 03:33 | <zewt> | site:whatwg.org/specs gives me tons of links under http://www.whatwg.org/specs/web-apps/2009-10-27/, http://www.whatwg.org/specs/web-apps/current-work/.w3c-html-core/ and other noise now ... wish thsoe would either be blocked from search engines or moved out of /specs |
| 03:37 | <Hixie> | if you give me a robots.txt that blocks what you want to block, i'll happily add it |
| 03:47 | <zewt> | well, I have no idea what all of the stuff under there is, heh |
| 04:49 | <zewt> | i find it painfully discouraging that i actually have to spend time arguing for not baking manual word-wrapping into a text format in 2012 |
| 04:51 | <zewt> | feels like i'm trying to argue a language designer out of putting line numbers on every line and having GOTO N be the primary form of flow control |
| 05:26 | <WeirdAl> | rniwa: ping on UndoManager spec... what DOM mutation events / observer notifications are required to fire on undoing an action? |
| 05:26 | <WeirdAl> | DOMAttrModified & friends |
| 05:26 | <rniwa> | WeirdAl: hi. yes. |
| 05:27 | <WeirdAl> | :) I'm wondering if they're all required |
| 05:27 | <rniwa> | WeirdAl: mutations made by the undo manager are regular DOM mutations |
| 05:27 | <rniwa> | WeirdAl: they're. |
| 05:27 | <rniwa> | WeirdAl: however, DOM mutation events are deprecated API so I can be careless. |
| 05:27 | <WeirdAl> | mutation observers are replacing them |
| 05:27 | <WeirdAl> | so, no, you can't ;) |
| 05:27 | <rniwa> | WeirdAl: they should certainly be included in the mutation records that mutation observers receive |
| 05:29 | <WeirdAl> | Should there be at least a token mention about mutation observers in section 3.1? |
| 05:30 | WeirdAl | is planning on implementing a "partial DOM" which is non-compliant in many respects, but UndoManager he'll probably want to fully implement |
| 05:33 | <rniwa> | WeirdAl: that makes sense. |
| 05:33 | <rniwa> | WeirdAl: jsdom? |
| 05:33 | <rniwa> | WeirdAl: or dom.js? |
| 05:33 | <WeirdAl> | neither |
| 05:33 | <rniwa> | oh, i see. |
| 05:34 | <WeirdAl> | and no, not envjs |
| 05:34 | <WeirdAl> | I'm going off the deep end... I've been forced to conclude that I pretty much have to write my own for a special purpose |
| 05:34 | rniwa | thinking maybe I can use WeirdAl's partial dom for https://plus.google.com/105748986001435560355/posts/aDV61jgSNXj |
| 05:35 | <WeirdAl> | forget it, you won't want it :) |
| 05:35 | <rniwa> | WeirdAl: ? |
| 05:35 | <WeirdAl> | it won't be implementing HTML :D |
| 05:35 | <rniwa> | WeirdAl: I see. |
| 05:35 | <rniwa> | that's interesting |
| 05:35 | <WeirdAl> | at least, not for probably a year or so |
| 05:36 | <WeirdAl> | I'm looking more at the XML world... it's cleaner :) |
| 05:36 | <rniwa> | I see. |
| 05:36 | <rniwa> | WeirdAl: but I hear that XML world is doomed |
| 05:36 | <WeirdAl> | maybe it is |
| 05:37 | <WeirdAl> | but I think that's because we just don't have good tools to edit it |
| 05:37 | <WeirdAl> | we have tools |
| 05:37 | <WeirdAl> | they're just not that good at it |
| 05:38 | <WeirdAl> | more specifically, the XML languages we humans are most likely to edit - XHTML, MathML, SVG, XUL, XBL, etc. - those are the ones for which the tools frankly need a lot of work |
| 05:40 | <rniwa> | true |
| 05:40 | <rniwa> | WeirdAl: I want a good editor for MathML. |
| 05:41 | <WeirdAl> | best I've seen for free is Amaya, and it's rather painful to work with |
| 05:41 | <rniwa> | yeah... |
| 05:41 | <WeirdAl> | believe me, I'm working on a new kind of XML editor in my spare time, one where new XML languages are literally like Firefox addons |
| 05:42 | <WeirdAl> | but to get there, I'm trying to build the tools to build those addons |
| 05:46 | <zcorpan> | Hixie: the status box in http://www.whatwg.org/specs/web-apps/current-work/multipage/editing.html#editing-0 looks like a bogus edit |
| 05:51 | <rniwa> | WeirdAl: i'll look forward to it :) |
| 05:51 | <Hixie> | zcorpan: feel free to fix it :-) |
| 05:59 | <zcorpan> | Hixie: i don't know what the previous state was... but, changed to something less inaccurate |
| 05:59 | <Hixie> | thanks |
| 07:45 | <annevk> | I wish the source code of the spider from http://dotnetdotcom.org/ was available so we could create a fresh index |
| 08:11 | <annevk> | what's the best way to represent an ordered dictionary in JSON? |
| 08:12 | <annevk> | it's about mapping one offset to another |
| 08:13 | <annevk> | e.g. 0:80, 36:A5, 50:B8, .. |
| 08:13 | <annevk> | nested array? |
| 08:13 | <annevk> | seems kind of ugly |
| 08:14 | <othermaciej> | array containing (the dictionary, an array of the keys in order) |
| 08:17 | <annevk> | I guess since the keys are sortable I could also do that during lookup... |
| 08:19 | <annevk> | kind of annoying that only gb18030 needs this special kind of index whereas all other encodings can do with a simple index |
| 08:20 | <annevk> | China, you're annoying! |
| 08:22 | <annevk> | back later |
| 08:46 | <david_carlisle> | rniwa: what kind of environment are you looking for an editor? personally I have a biased view on the worls and use emacs for everything, but firefox addon http://www.maths-informatique-jeux.com/blog/frederic/?post/2010/11/14/Mozilla-MathML-Add-ons is quite promising, or shockingly enough, Word isn't bad, there are several others it depends what you want... |
| 09:03 | <Ms2ger> | Well done, public-webapps! |
| 09:04 | <Ms2ger> | January to March 2012 ... 1337 messages |
| 09:04 | <annevk> | nice |
| 09:11 | <annevk> | since the gb18030 index is so different from other indexes, should I give it a different name? |
| 09:11 | <annevk> | and if so, suggestions? |
| 09:22 | <MikeSmith> | Ms2ger: can you remind me, was somebody working already on testharness.js-enabling the canvas test suite? |
| 09:22 | <Ms2ger> | Indeed |
| 09:22 | <Ms2ger> | I was! |
| 09:22 | <Ms2ger> | It's in a bug somewhere |
| 09:22 | <MikeSmith> | oh cool |
| 09:23 | <Ms2ger> | Waiting for Philip` |
| 09:23 | <MikeSmith> | w3c bug? |
| 09:23 | <Ms2ger> | Yep |
| 09:23 | <MikeSmith> | oh |
| 09:23 | <MikeSmith> | waiting on Philip` to just land your changes? |
| 09:23 | <Ms2ger> | To see if it makes sense to him |
| 09:23 | <MikeSmith> | OK |
| 09:24 | MikeSmith | looks for the bug |
| 09:24 | <MikeSmith> | found it |
| 09:24 | <MikeSmith> | https://www.w3.org/Bugs/Public/show_bug.cgi?id=14191 |
| 09:25 | <zcorpan> | annevk: index-gb18030-Y-U-DIFFERENT???.txt |
| 09:29 | <annevk> | heh |
| 09:29 | <annevk> | so the way gb18030 works is that there are 207 ranges |
| 09:30 | <annevk> | and by computing a number from the four bytes sequences |
| 09:30 | <annevk> | the ranges are basically offsets |
| 09:31 | <annevk> | consisting of "offset, code point offset" |
| 09:31 | <annevk> | you then find the last range whose offset is equal or less than the computed index |
| 09:31 | <annevk> | and then you do computed index - offset + code point offset |
| 09:31 | <annevk> | and you have a code point |
| 09:32 | <annevk> | on top of that you exclude computed indexes between 39419 and 189000 and anything greater than 1237575 |
| 09:32 | <MikeSmith> | the webgl test runner is nice |
| 09:32 | <MikeSmith> | https://cvs.khronos.org/svn/repos/registry/trunk/public/webgl/sdk/tests/webgl-conformance-tests.html |
| 09:33 | <annevk> | so I guess I'll just call it index-gb18030 like the others and explain in prose look works differently |
| 09:33 | <annevk> | and in JSON I'll store it as an object of which you need to sort the keys yourself |
| 09:33 | <annevk> | and then publish the JSON for the indexes as indexes.json |
| 09:34 | <annevk> | and publish the index.py which generates the index-*.txt files |
| 09:34 | <annevk> | merge the single-byte encodings into it |
| 09:34 | <annevk> | and have a separate encodings.json which lists the encoding names and labels |
| 09:38 | <annevk> | http://i.imgur.com/vynW8.png o_O |
| 09:39 | <MikeSmith> | david_carlisle: btw, is MathML support still not enabled in Chrome? |
| 09:39 | <MikeSmith> | annevk: heh |
| 09:39 | <charlvn> | annevk: it's a pity we can't see the nationality of those people on the pic... i bet i know what the results will be :) |
| 09:41 | <Ms2ger> | Dutch? :) |
| 09:41 | <MikeSmith> | david_carlisle: OK, I see that it's still not |
| 09:41 | <Ms2ger> | MikeSmith, the webgl test harness, otoh, sucks badly |
| 09:41 | <david_carlisle> | MikeSmith: No, but there is a new person working on the mathml in the webkit codebase and google people are in teh reviewing loop this time, so things are looking better |
| 09:41 | <MikeSmith> | Ms2ger: yeah, I got about 10% through it and it crashes ever time |
| 09:41 | <Ms2ger> | Blame your graphics driver |
| 09:42 | <MikeSmith> | crashes in every browser I have tried with so far |
| 09:42 | <MikeSmith> | david_carlisle: given that it's enabled in Safari, I wonder why not in Chrome. Is the issue that it's too incomplete at this point? |
| 09:43 | <Ms2ger> | Because Chrome has *some* quality standards? |
| 09:43 | <charlvn> | Ms2ger: i had an idea of something a bit further west ;) |
| 09:43 | <MikeSmith> | david_carlisle: pages like http://www.mozilla.org/projects/mathml/demo/texvsmml.html seem to mostly render as expected |
| 09:43 | <MikeSmith> | in Safari |
| 09:43 | <david_carlisle> | MikeSmith: The stated reason is that it did not have relevant security review, whether there was actual security concerns or if that is just teh trump card to play to avoid doing anything I am not in a position to say:-0 |
| 09:44 | <MikeSmith> | ah, OK |
| 09:44 | <Ms2ger> | annevk, hmm, fascinating how you were just looking at that page :) |
| 09:45 | <annevk> | recent changes to the WHATWG Wiki is one of the few speed dial things I have |
| 09:45 | <annevk> | so sometimes when I open a new tab I click it |
| 09:46 | <MikeSmith> | Ms2ger: found in source of WebGL test page: <script src="../../resources/js-test-pre.js"/> |
| 09:46 | <MikeSmith> | self-closing tag |
| 09:46 | <Ms2ger> | Yeah |
| 09:46 | <Ms2ger> | They come from webkit |
| 09:46 | <MikeSmith> | not a good sign |
| 09:47 | <zcorpan> | webkit still does that? |
| 09:49 | <MikeSmith> | hmm, that test runner doesn't think WebKit nightly supports WebGL |
| 09:49 | <annevk> | is |
| 09:49 | <annevk> | "this index works different from ..." |
| 09:49 | <annevk> | or "this index works differently from ..." |
| 09:49 | <Ms2ger> | than? ;) |
| 09:50 | <annevk> | This index works differently from all others as ... |
| 09:50 | <annevk> | This index works differently than all others as ... |
| 09:50 | <annevk> | ? |
| 09:50 | <annevk> | don't think so |
| 09:50 | <annevk> | well maybe |
| 09:52 | MikeSmith | finds he has to do defaults write com.apple.Safari WebKitWebGLEnabled -bool YES |
| 09:56 | <MikeSmith> | fwiw, this is the crashy test: https://cvs.khronos.org/svn/repos/registry/trunk/public/webgl/sdk/tests/conformance/context/context-creation-and-destruction.html |
| 09:56 | <MikeSmith> | WebKit is the only browser I've tried that it doesn't crash in |
| 10:00 | <MikeSmith> | Ms2ger: btw a couple of your apis-in-html-documents tests crash validator.nu when I try to validate them |
| 10:01 | <Ms2ger> | Nice |
| 10:01 | <MikeSmith> | figured out the fix already |
| 10:01 | <MikeSmith> | http://bugzilla.validator.nu/show_bug.cgi?id=917 |
| 10:01 | <Ms2ger> | Also, is http://w3c-test.org/shepherd/ ever going to do something else than print an error message? |
| 10:01 | <MikeSmith> | talk to Berjon |
| 10:02 | <MikeSmith> | I didn't set that up |
| 10:02 | <MikeSmith> | and I'm not supposed to work on the framework code these days |
| 10:02 | <Ms2ger> | :( |
| 10:09 | <annevk> | Ms2ger: http://www.answers.com/topic/different |
| 10:10 | <zcorpan> | MikeSmith: messages about attributes on <input> really needs improving it seems |
| 10:11 | <charlvn> | lol: http://bugs.jquery.com/ticket/10891 |
| 10:11 | <MikeSmith> | zcorpan: so last time I thought about how to solve that the thought I remember having is that it'd be best to do it in the JS for the Web UI |
| 10:12 | <Ms2ger> | Oh, lovely |
| 10:12 | <Ms2ger> | if ( !window.opera ) { |
| 10:12 | <MikeSmith> | zcorpan: why do you mention it? somebody raised a new bug? |
| 10:12 | <Ms2ger> | This ringmark thing does seem like a good idea |
| 10:12 | <gsnedders> | Ms2ger: Why wouldn't you do browser sniffing? |
| 10:12 | <zcorpan> | MikeSmith: someone on help@whatwg |
| 10:13 | <MikeSmith> | ok |
| 10:13 | <Ms2ger> | I was already assuming it was useless |
| 10:13 | <Ms2ger> | Now, I'm certain |
| 10:13 | <MikeSmith> | zcorpan: I can try to write up something for it this week. |
| 10:14 | <zcorpan> | MikeSmith: cool |
| 10:14 | <hsivonen> | zcorpan: nice IE behavior |
| 10:15 | <hsivonen> | Ms2ger: ringmark checks for window.opera? |
| 10:15 | <Ms2ger> | Yep |
| 10:15 | <zcorpan> | hsivonen: ? |
| 10:15 | <hsivonen> | zcorpan: the <input name="nodeName"> thing |
| 10:16 | <hsivonen> | Ms2ger: :-( |
| 10:16 | <zcorpan> | hsivonen: i'm missing context :-) |
| 10:16 | <gsnedders> | Ms2ger: Where is the sniffing? |
| 10:17 | <Ms2ger> | http://rng.io/dist/rings.js |
| 10:18 | <charlvn> | the nodeName thing is strange because i had a similar problem and ended up hacking jquery too in a slightly different way https://gist.github.com/2350137 |
| 10:19 | <charlvn> | (i am stuck with using some third party libs) |
| 10:19 | <hsivonen> | zcorpan: oops. I misread who said what. the jquery link was given by charlvn |
| 10:19 | <zcorpan> | ah |
| 10:28 | <annevk> | teehee |
| 10:28 | <annevk> | all encodings now have a defined decoder algorithm |
| 10:28 | <annevk> | including gb18030 |
| 10:29 | <annevk> | of course, there's a few bugs |
| 10:29 | <annevk> | but still |
| 10:45 | <zcorpan> | nice |
| 10:57 | <bga_> | gsnedders in Opera new Date(NaN).toJSON() throw RangeError but according spec it should return null |
| 10:59 | <gsnedders> | toJSON? |
| 11:00 | <gsnedders> | Oh, that does exist, but only on Date. |
| 11:42 | <annevk> | david_carlisle: btw, I'm pretty close to having a single JSON file that has the indexes for all encodings |
| 11:42 | <annevk> | david_carlisle: along the lines you suggested ages ago |
| 11:43 | <annevk> | david_carlisle: it took me a while to get the proper separation sorted out, as I didn't read up much history and just reverse engineered everything :) |
| 11:44 | <david_carlisle> | annevk: any more jobs I could give you?;-) |
| 11:49 | <annevk> | david_carlisle: heh, if they're as fascinating as this one, sure |
| 11:49 | <annevk> | this standard is probably the coolest thing I've worked on to date |
| 11:50 | <zcorpan> | cooler than css namespaces? :-) |
| 11:51 | zcorpan | wishes window.documetn was defined and returned window.document |
| 11:51 | <zcorpan> | i always make that typo |
| 11:55 | <annevk> | zcorpan: I'm not sarcastic, I really like working on the Encoding Standard |
| 11:56 | Ms2ger | puts up the sarcasm sign anyway |
| 11:57 | <annevk> | heh |
| 11:57 | <annevk> | I mean, if I didn't, why would I be doing it? |
| 11:58 | <zcorpan> | oh, i didn't think you were being sarcastic |
| 11:58 | <zcorpan> | anyway, document.head.outerHTML = 'x' is weird |
| 11:59 | <Ms2ger> | Fascinating |
| 12:00 | <zcorpan> | or '<link>x' |
| 12:00 | <Ms2ger> | zcorpan, are you writing tests? :) |
| 12:00 | <zcorpan> | i was gonna write tests for outerHTML in xml, but stumbles upon this thing when intending to check what happened when setting outerHTML on root in opera |
| 12:01 | <zcorpan> | s/s/d/ |
| 12:12 | <hsivonen> | https://www.w3.org/2002/09/wbs/40318/issue-158-objection-poll/results looks like a success for the Decision Policy. </sarcasm> |
| 12:13 | <zcorpan> | LOL |
| 12:15 | <Ms2ger> | That's pretty much the best case scenario in terms of wasting people's time |
| 12:17 | <hsivonen> | hmm. LLVM is much larger than I thought |
| 12:18 | <hsivonen> | in terms of number of source files and compilation time |
| 13:08 | <MikeSmith> | annevk: congrats on getting all the encodings defined |
| 13:08 | <annevk> | now the encoders... |
| 13:08 | <annevk> | and the bugs |
| 13:08 | <MikeSmith> | and the test suite |
| 13:09 | <annevk> | you keep thinking you're at 80 percent, but it's barely 8 |
| 13:14 | <zcorpan> | Ms2ger: http://html5.org/specs/dom-parsing.html#concept-serialize-xml - instead of throwing for ]]> in text node with serializeAsCDATA, i think you should serialize a cdata section up to the first "]]>" not including the ">", then emit the rest (including the ">") as text |
| 13:14 | <Ms2ger> | File a bug? :) |
| 13:15 | <zcorpan> | done |
| 13:22 | <annevk> | can't believe that in 2012 I have to write this in Python |
| 13:22 | <annevk> | def char(cp): |
| 13:22 | <annevk> | if cp > 0xFFFF: |
| 13:22 | <annevk> | hi, lo = divmod(cp-0x10000, 0x400) |
| 13:22 | <annevk> | return unichr(0xD800+hi) + unichr(0xDC00+lo) |
| 13:22 | <annevk> | return unichr(cp) |
| 13:22 | <annevk> | but then you need to do the same in JavaScript I guess |
| 13:23 | <annevk> | divmod is a pretty neat built-in though |
| 13:23 | <annevk> | david_carlisle: http://dvcs.w3.org/hg/encoding/raw-file/tip/indexes.json |
| 13:24 | <annevk> | normative indexes look a little prettier now |
| 13:24 | <annevk> | http://dvcs.w3.org/hg/encoding/raw-file/tip/index-big5.txt |
| 13:26 | <annevk> | where if you do some copy/pasting/searching you'll find browsers have many embarrassing font issues |
| 13:28 | <david_carlisle> | annevk: lovely! would be lovelier still in xml of course (but joking aside, looks good and .txt version looks nice too) |
| 13:28 | <annevk> | XML haha |
| 13:28 | <annevk> | they tried that |
| 13:28 | <annevk> | it got way complex |
| 13:29 | <annevk> | to actually get the simple gb18030 mappings I parsed some XML with a regular expression :p |
| 13:33 | <gsnedders> | jgraham: Seem to be going for your recommendation, so thanks! |
| 13:34 | <annevk> | zcorpan: you could also do (WindowProxy or MessagePort)? |
| 13:34 | <annevk> | zcorpan: might be cleaner |
| 13:34 | <annevk> | zcorpan: especially since these can be constructed via dictionary where you'll need to do some type checking |
| 13:38 | <david_carlisle> | annevk: do you need the double negative here "A byte never consists of more, nor less than exactly eight bits" or is this just "A byte consists of exactly eight bits" (if there is a subtle distinction. could you make it less subtle:-) |
| 13:40 | <zcorpan> | annevk: oh, didn't know that worked for return values |
| 13:40 | <annevk> | david_carlisle: fair enough |
| 13:40 | <Tuju> | i've problem with xml namespace and getting xslt to match its elements |
| 13:41 | <Tuju> | if i remove all xmls etc attributes from xml root tag, it works, but with them it doesn't. |
| 13:41 | <Tuju> | how do i specify the namespace into xslt ? |
| 13:42 | <annevk> | use xmlns:prefix=namespace |
| 13:42 | <annevk> | and then use prefix in the XPath expressions? |
| 13:42 | <annevk> | dunno though, it's been at least half a decade since I played with XSLT |
| 13:43 | <Tuju> | hmmm |
| 13:43 | <david_carlisle> | Tuju: as annevk said, or if you are using xslt2 set xpath-default-namespace then you don't need a prefix (this is the wrong channel for xslt questions though, no one here except me likes xml at all :-( |
| 13:43 | <Tuju> | annevk: i think you're in something, since after googling i saw something similar too, but haven't really found any good example. |
| 13:44 | <zcorpan> | hmm. i checked the first search result ddg gave me for xslt namespace prefix. opera crashed. clearly opera doesn't want me to spend time reading about this subject |
| 13:44 | <annevk> | david_carlisle: robbertbroersma does, but he's not here right now |
| 13:45 | <david_carlisle> | annevk: I rest my case:-) |
| 13:52 | <MikeSmith> | all the git docs are generated using xslt, including the man pages / "git help foo" output |
| 13:52 | <MikeSmith> | take a look at the source of e.g. /usr/share/man/man1/git-pull.1.gz |
| 13:53 | <MikeSmith> | I know because I wrote the xslt that's used to generate them |
| 13:53 | <MikeSmith> | that said, I still don't count as somebody who likes xslt |
| 13:56 | <annevk> | foolip: you around? |
| 13:56 | <foolip> | annevk, yep |
| 13:57 | <annevk> | foolip: fwiw, the index data is now in the spec |
| 13:57 | <annevk> | foolip: I'm gonna take a look again at big5 now |
| 13:57 | <annevk> | foolip: to make it up to date with the thread |
| 13:57 | <foolip> | annevk, which index data, the two missing points? |
| 13:58 | <annevk> | foolip: sorry, the JSON file is now in the repository |
| 13:58 | <foolip> | annevk, ah |
| 13:58 | <foolip> | good |
| 13:59 | <annevk> | foolip: I have not updated the spec yet with respect to the big5 changes suggested as I lost track of what was being suggested after Γistein joined |
| 13:59 | <foolip> | annevk, I emailed public-html-ig-zhβwo asking for some souble checking: http://lists.w3.org/Archives/Public/public-html-ig-zh/2012Apr/0001.html |
| 13:59 | <foolip> | useful reply (in English) at http://lists.w3.org/Archives/Public/public-html-ig-zh/2012Apr/0002.html |
| 13:59 | <foolip> | it looks like at least one of the pages I looked at depends on the Firefox Big5 extensions, in other words that user must have posted using Firefox |
| 14:00 | <foolip> | however, AFAICT there's nothing that can be done to save it |
| 14:00 | <foolip> | annevk, I'll wait for a few more days to see if more replies come in and then summarize the changes that I agree should be made |
| 14:01 | <foolip> | It'll be fun to see if GBK and GB18030 are just as messy |
| 14:01 | <Ms2ger> | "fun" |
| 14:01 | <foolip> | BTW, I think we should merge them as well, we're bound to find pages labeled as GBK which are in fact GB18030 |
| 14:01 | <annevk> | gbk and gb18030 were easy |
| 14:01 | <annevk> | apart from a flag they're effectively merged |
| 14:01 | <annevk> | see the spec |
| 14:02 | <foolip> | I've seen it, and the note |
| 14:02 | <foolip> | Is there a downside to not having the flag? |
| 14:02 | <annevk> | I removed the flag for the euro sign already |
| 14:02 | <annevk> | dunno about doing it for the rest |
| 14:02 | <annevk> | I guess we could |
| 14:02 | <foolip> | what does IE do? |
| 14:03 | <annevk> | but I don't think any implementation has them merged at this point |
| 14:03 | <foolip> | annevk, have you tried to search for pages labeled as GBK that would decode with less errors if treated as GB18030? |
| 14:04 | <annevk> | no I haven't |
| 14:04 | <annevk> | I've just checked what browsers did today and specced that, minus PUA |
| 14:04 | <foolip> | OK, fair enough |
| 14:04 | <annevk> | it's easy to merge though and I've no objections if implementors want to do that |
| 14:05 | <foolip> | I'm interested in doing some Big5 tests, is it http://dotnetdotcom.org/ I should be using? |
| 14:05 | <annevk> | we'd need to decide on whether to call it gbk or gb18030 and that's about it |
| 14:05 | <annevk> | foolip: you can only use that to find URLs |
| 14:05 | <annevk> | foolip: you then need to fetch the URLs yourself because their data is UTF-8 normalized |
| 14:05 | <annevk> | foolip: I'll wait with updating big5 then if you're going to post a summary later |
| 14:06 | <foolip> | annevk, do you already have the index file so that you could easily send me a list of URLs that claim to be big5* and gb*? |
| 14:06 | <annevk> | no zcorpan does, but last time I had to extract the URLs out of that data myself |
| 14:06 | <annevk> | I should still have that extract URLs script though |
| 14:07 | <foolip> | that would be useful, yeah |
| 14:08 | <annevk> | urls = [] |
| 14:08 | <annevk> | url = "" |
| 14:08 | <annevk> | state = "start" |
| 14:08 | <annevk> | for b in open("big5-with-urls.txt", "rb").read(): |
| 14:08 | <annevk> | if state == "start" and b == "\x0A": |
| 14:08 | <annevk> | state = "url" |
| 14:08 | <annevk> | elif state == "url": |
| 14:08 | <annevk> | if b in "\x0A\x0D": |
| 14:08 | <annevk> | state = "start" |
| 14:08 | <annevk> | url = "" |
| 14:08 | <annevk> | elif b == "\x00": |
| 14:08 | <annevk> | state = "appendurl" |
| 14:08 | <annevk> | urlfile.write(url + "\n") |
| 14:08 | <annevk> | big5-with-urls.txt was the result of a grep zcorpan did |
| 14:08 | <foolip> | zcorpan, do you have big5-with-urls.txt or the script to produce it? |
| 14:09 | <annevk> | oh that script is not complete |
| 14:09 | <annevk> | I'll email it to www-archive |
| 14:09 | <zcorpan> | http://simon.html5.org/dump/big5-with-urls.txt.zip ? |
| 14:10 | <foolip> | zcorpan, would it be possible for you to grep me a gb-with-urls.txt.zip as well? |
| 14:10 | <zcorpan> | should i do a grep for gb* ? |
| 14:10 | <foolip> | that would be great, yes |
| 14:10 | <foolip> | are you catching both Content-Type header and <meta charset>? |
| 14:10 | <zcorpan> | yep |
| 14:11 | <foolip> | cool |
| 14:11 | <zcorpan> | do you want all the labels under "Legacy multi-byte Chinese (simplified) encodings" in the encoding spec? |
| 14:12 | <foolip> | zcorpan, chinese, csgb2312, ... yeah, I guess that's best |
| 14:12 | <foolip> | although I've never seen half of them |
| 14:14 | <zcorpan> | LANG=C grep -aEizB2 "(content(-type[[:space:]]*:|=[[:space:]]*[\"']?)[[:space:]]*text/html[[:space:]]*;|<meta[[:space:]])[[:space:]]*charset[[:space:]]*=[[:space:]]*[\"']?(chinese|csgb2312|csiso58gb231280|gb2312|gb_2312|gb_2312-80|gbk|iso-ir-58|x-gbk|gb18030|hz-gb-2312)" web200904 > gb-with-urls.txt |
| 14:15 | <zcorpan> | ping me in a few minutes and i'll upload it |
| 14:15 | <foolip> | zcorpan, is that dump from 2009 as the filename would suggest? |
| 14:16 | <annevk> | yes |
| 14:16 | <zcorpan> | yep |
| 14:16 | <foolip> | so basically dotnetdotcom.org is dead? |
| 14:16 | <annevk> | probably |
| 14:16 | <foolip> | too bad archive.org doesn't have some kind of index or "random page" function :( |
| 14:17 | <annevk> | I was saying earlier today it would be great to have a new source of data |
| 14:17 | <annevk> | per http://whois.domaintools.com/dotnetdotcom.org they have been maintaining their registration at least |
| 14:22 | <zcorpan> | that ended up being 862MB (before zipping) -- lemme run the python script |
| 14:26 | zcorpan | needs to free memory |
| 14:35 | <annevk> | Push SMS again |
| 14:36 | <zcorpan> | wow the python script is slow |
| 14:36 | <annevk> | well, you just fed it an order of magnitude more data than it worked on originally |
| 14:37 | <annevk> | prolly needs some time to get adjusted to these harsh conditions |
| 14:37 | <zcorpan> | i think it finished now |
| 14:37 | <zcorpan> | after about 7 minutes of cpu time |
| 14:37 | <annevk> | I'm surprised that actually worked |
| 14:38 | <annevk> | most have used a hell of a lot of memory |
| 14:39 | <zcorpan> | foolip: http://simon.html5.org/dump/gb-urls.txt.zip |
| 14:39 | <zcorpan> | annevk: 8GB, though the script didn't seem to use more than 1 |
| 14:40 | <foolip> | zcorpan, great, so that's just a list of the URLs, then? |
| 14:40 | <zcorpan> | foolip: yep |
| 14:40 | <foolip> | thanks, I'll have a look at it tonight! |
| 14:41 | <annevk> | zcorpan: seems about right |
| 14:41 | <annevk> | zcorpan: it reads the whole fill into memory |
| 14:41 | <zcorpan> | yeah |
| 14:41 | <annevk> | zcorpan: that's probably what takes a long time |
| 14:41 | <annevk> | zcorpan: reading half a megabyte of encoding data is already slowish |
| 14:42 | <annevk> | well, reading and parsing as JSON |
| 14:43 | <zcorpan> | i found a python script that defined an iterator that could split on zero bytes the other day, but then found i didn't need it. i guess i should look for it again, it'll probably be useful later |
| 14:50 | <zcorpan> | maybe something from http://bytes.com/topic/python/answers/41987-canonical-way-dealing-null-separated-lines |
| 14:50 | <zcorpan> | but that doesn't have the indentation |
| 15:04 | <Tuju> | annevk: i still can't get that namespace to work |
| 15:04 | <Tuju> | it's tricky |
| 15:06 | <annevk> | Tuju: on xsl:stylesheet you declare xmlns:x="yournamespace" |
| 15:06 | <Tuju> | annevk: what i read, plain xmlns="example.com" would do default namespace, right? |
| 15:06 | <annevk> | Tuju: then in e.g. xsl:template you use match="//x:myElement" |
| 15:07 | <Tuju> | and then i would avoid writing that x: everywhere. |
| 15:07 | <annevk> | Tuju: no that's not how it works |
| 15:07 | <Tuju> | but it works in examples that they've used in many pages |
| 15:07 | <Tuju> | they don't write that x: in every xslt tag. |
| 15:07 | <annevk> | in XSLT? |
| 15:07 | <Tuju> | yup yup |
| 15:07 | <annevk> | maybe they're using XPath 2.0 |
| 15:07 | <annevk> | dunno about that |
| 15:08 | <MikeSmith> | Tuju: http://www.mulberrytech.com/xsl/xsl-list/ is a good place to get xslt help |
| 15:08 | <Tuju> | http://linux.dd.com.au/wiki/XSLT_Tutorial for example that one |
| 15:08 | <zcorpan> | Tuju: if it worked when you removed xmlns, it means it matched against the no namespace |
| 15:09 | <Tuju> | annevk: that even doesn't declare the default one. |
| 15:09 | <Tuju> | i don't mind if i have one if it's been fixed once as default and doesn't need to be written into every tag. |
| 15:10 | <MikeSmith> | Tuju: that example input document is not in any namespace man |
| 15:10 | <MikeSmith> | that's why the stylesheet doesn't need to use namespace prefixes |
| 15:11 | <MikeSmith> | maybe you should just remove the namespace declarations from whatever your input source is |
| 15:11 | <Tuju> | MikeSmith: but if my input doc has a namespace, can't i just set it into default in my xslt? |
| 15:11 | <MikeSmith> | not in xslt 1.0 you can't, no |
| 15:11 | <MikeSmith> | or xpath 1.0 |
| 15:12 | <Tuju> | if i just do one namespace declaration in input doc, i'd like to do same way in xslt too. |
| 15:12 | <Tuju> | MikeSmith: so there is no way around having that x: (orwhatever) in every god damn tag then? |
| 15:12 | <MikeSmith> | yeah there is |
| 15:12 | <MikeSmith> | make your input source not be in any namespace |
| 15:13 | <Tuju> | yes, but i rather not to do that. |
| 15:13 | <MikeSmith> | yeah well |
| 15:13 | <MikeSmith> | life is full of choices |
| 15:13 | <Tuju> | is there a way to get around it with xsltproc options? |
| 15:13 | <MikeSmith> | blame xml |
| 15:13 | <Tuju> | i don't want to, not yet. :) |
| 15:14 | <MikeSmith> | Tuju: no way to work around it with options in xsltproc no |
| 15:14 | <Tuju> | you ruined my day :) |
| 15:14 | <Tuju> | with facts. |
| 15:14 | <Tuju> | maybe i start writing those damn x:'s then... |
| 15:15 | <MikeSmith> | I could actually be wrong |
| 15:15 | <Tuju> | xml - solution if you really want to write chars a lot... |
| 15:15 | <MikeSmith> | you should not just trust anything I say about xslt |
| 15:16 | <Tuju> | MikeSmith: so far the empiric evidence implies that you're right. |
| 15:16 | <MikeSmith> | as david_carlisle pointed out, this is the last places you should turned to for quality help with xslt |
| 15:17 | <Tuju> | :) |
| 15:17 | <Tuju> | i'm afraid that this is the best in irc |
| 15:22 | <MikeSmith> | annevk: so what more needs to be defined about directionality in the Notifications spec? |
| 15:24 | <annevk> | general FYI: I changed terminology around indexes slightly |
| 15:24 | <annevk> | an index now is a list of pointers and corresponding code points |
| 15:25 | <annevk> | MikeSmith: there was a thread on that on the list |
| 15:25 | <annevk> | MikeSmith: I haven't really taken the time to study that yet |
| 15:25 | <MikeSmith> | ok |
| 15:25 | <annevk> | MikeSmith: in addition, there's quite a few flaws with the draft |
| 15:25 | <MikeSmith> | anyway, I pinged Richard about it |
| 15:26 | <annevk> | MikeSmith: I think the i18n guys raised it |
| 15:26 | <MikeSmith> | oh |
| 15:26 | <annevk> | MikeSmith: e.g. just says "object" instead of defining a dictionary |
| 15:26 | <annevk> | MikeSmith: doesn't use the proper HTML terminology |
| 15:26 | <annevk> | MikeSmith: I wonder if I should take some time to rewrite it using Anolis |
| 15:26 | <annevk> | MikeSmith: and fix most obvious bugs |
| 15:28 | <david_carlisle> | annevk: you should write it in xmlspec xml and process it with xslt, You can always ping MikeSmith for any xslt help required |
| 15:30 | <MikeSmith> | I learned all the xslt I know from reading and copying and hacking Norm Walsh's code |
| 15:30 | <MikeSmith> | annevk: you dudes need to get over your obsession with anolis |
| 15:30 | <annevk> | david_carlisle: so much trolling :) |
| 15:30 | <annevk> | MikeSmith: there's nothing better |
| 15:31 | <annevk> | maybe a hybrid of anolis/respec would be, but that doesn't exist |
| 15:31 | <MikeSmith> | let's fix respec, if it's not doing what we need |
| 15:31 | <MikeSmith> | pytho-reactionaries living in the past man |
| 15:31 | <MikeSmith> | 2012 is a JS world |
| 15:32 | <Ms2ger> | So, first, make it work so that it works with JS disabled |
| 15:32 | <annevk> | it's more that between writing specs and writing a tool to write specs, I prefer the former |
| 15:32 | <annevk> | also, I use Python a lot still to process data and generate things |
| 15:33 | <MikeSmith> | I prefer getting somebody else to both write specs for me and write tools for writing specs |
| 15:33 | <annevk> | I guess if a v8 shell or some such came installed I could switch some of that, but not sure if it's worth it |
| 15:33 | <Ms2ger> | Spidermonkey! ;) |
| 15:33 | <annevk> | MikeSmith: that's why you're a pimp :p |
| 15:34 | <MikeSmith> | heh |
| 15:35 | <annevk> | I think I agree with you in spirit, but respec does not do what I need, and anolis does, apart from it being slightly harder to pick up for newbies |
| 15:35 | <annevk> | and every spec that is generated with respec I sort of feel should be redone some day... |
| 15:36 | <annevk> | its reverse IDL'ing and such is just wrong |
| 15:36 | <MikeSmith> | yeah, agreed about that |
| 15:37 | <MikeSmith> | annevk: anyway, in other news I met with Martin Duerst the other day to talk about the URL spec and Encoding spec |
| 15:38 | <MikeSmith> | and he was generally not strongly unfavorable to both |
| 15:38 | <annevk> | not strongly unfavorable :) |
| 15:38 | <MikeSmith> | heh |
| 15:38 | <MikeSmith> | yeah |
| 15:40 | <MikeSmith> | also btw, the IRI WG is not going to take any position on the web+ scheme |
| 15:40 | <annevk> | how is URL going? |
| 15:40 | <MikeSmith> | it's not |
| 15:41 | <MikeSmith> | I don't want to work on it alone |
| 15:41 | <annevk> | lol http://dvcs.w3.org/hg/url/rev/dc617a4934cc |
| 15:41 | <karlcow> | http://pyjs.org/#Translator |
| 15:41 | <annevk> | MikeSmith: k |
| 15:41 | <annevk> | MikeSmith: it's top of my post-Encoding todo list |
| 15:41 | <Ms2ger> | MikeSmith, hah, you think that will work? |
| 15:41 | <MikeSmith> | :) |
| 15:41 | <Ms2ger> | Unless Hixie hires someone to do it :) |
| 15:42 | <Ms2ger> | That's the only reason I managed to get rid of DOM Range |
| 15:42 | <annevk> | I sort of have it in for the low-level stuff nowadays |
| 15:42 | <annevk> | Encoding, URL |
| 15:42 | <annevk> | MIME? |
| 15:42 | <karlcow> | url seems to be dead living draft |
| 15:42 | <annevk> | but, no rush |
| 15:43 | <MikeSmith> | annevk: mime sniffing! |
| 15:43 | <MikeSmith> | Ms2ger: :) |
| 15:43 | <MikeSmith> | annevk: Chris Weber has aggregated the WebKit tests into the set of tests he already made. So when you get time, we can start looking at those tests and then go from there |
| 15:43 | <annevk> | MikeSmith: well that's done, but parsing Content-Type and media-type and such is not |
| 15:43 | <annevk> | MikeSmith: yeah |
| 15:43 | <annevk> | the encoding stuff is taking somewhat longer than expected |
| 15:44 | <MikeSmith> | annevk: somebody I know is agitating me about the fact that the HTML spec references the mime-sniffing draft which is no longer being maintained as an IETF I.D. |
| 15:44 | <annevk> | blame Larry? |
| 15:44 | <MikeSmith> | um, er |
| 15:45 | <annevk> | "oh trololol i'm so experienced... lets split this all out... object object... tralala... to the IETF!" |
| 15:45 | <annevk> | happened to MIME sniff and URL |
| 15:45 | <annevk> | IETF is where HTML spec features go to die |
| 15:46 | <jwalden> | someone should send the DRM stuff over there, then |
| 15:46 | <Ms2ger> | jwalden++ |
| 15:46 | <MikeSmith> | hahaha |
| 15:46 | <MikeSmith> | people who actually do active work at IETF are great |
| 15:46 | <MikeSmith> | Peter St. Andre |
| 15:46 | <MikeSmith> | Chris Weber |
| 15:47 | <MikeSmith> | and such |
| 15:47 | jwalden | doesn't doubt that :-) |
| 15:47 | <Ms2ger> | "New public mailing list for discussions on potential NFC work at W3C" |
| 15:50 | <MikeSmith> | Mike Belshe seems to have deleted his tweet about the IETF from last week |
| 15:52 | <karlcow> | annevk: thanks for this. http://lists.w3.org/Archives/Public/www-archive/2012Apr/0034 |
| 15:52 | <karlcow> | I have been advocating this for a while but failed each time. :/ |
| 15:53 | <Ms2ger> | karlcow, I would suggest wiki.whatwg.org instead ;) |
| 15:53 | <karlcow> | Ms2ger: what is that? ;) |
| 15:53 | <Ms2ger> | A spam-infested hellscape |
| 15:54 | <annevk> | hmm http://commoncrawl.org/ |
| 15:56 | <annevk> | looks complicated to use |
| 15:56 | <annevk> | but promises access to billions of documents |
| 15:57 | <MikeSmith> | man some of the characters in http://dvcs.w3.org/hg/encoding/raw-file/tip/index-big5.txt are just crazy cool |
| 15:57 | <MikeSmith> | like π₯© |
| 15:58 | <MikeSmith> | or even simpler ones like η«Ό |
| 15:58 | <MikeSmith> | japanese doesn't have these |
| 15:59 | <annevk> | and resources are stored in http://archive.org/web/researcher/ArcFileFormat.php which is of course a spec written along the lines of HTML4 but then with more skiing and alcohol |
| 16:00 | <MikeSmith> | heh |
| 16:00 | <MikeSmith> | I hope you're not denigrating alcohol |
| 16:01 | <annevk> | only when combined with spec writing |
| 16:02 | <MikeSmith> | it makes everything better |
| 16:02 | <MikeSmith> | well almost everything, up to a point |
| 16:04 | <MikeSmith> | nose-painting, provokes, and unprovokes, etc. |
| 16:04 | <MikeSmith> | the Porter's speech |
| 16:05 | <MikeSmith> | stand to and not stand to |
| 16:07 | <annevk> | heh β‘ |
| 16:08 | <annevk> | monospace β § also falls apart |
| 16:16 | <TabAtkins> | Woo! http://www.w3.org/TR/css-variables/ |
| 16:17 | <annevk> | wrong channel to cheer about TR/ |
| 16:17 | <wilhelm> | Any implementations yet? |
| 16:17 | <TabAtkins> | wilhelm: Ours is in-progress. |
| 16:18 | <wilhelm> | Landing which year? (c: |
| 16:18 | <TabAtkins> | This year, hopefully. ^_^ |
| 16:18 | <wilhelm> | Cool. |
| 16:20 | <beverloo> | any wip patches around? |
| 16:21 | <beverloo> | there is one for hierarchies :) |
| 16:21 | <TabAtkins> | Variables is quite a bit more involved, so I dunno. Ask Luke, though. |
| 16:22 | <beverloo> | will do, thanks! |
| 16:25 | <MikeSmith> | um, webkit peoples, should you read the logs and find http://krijnhoetmer.nl/irc-logs/whatwg/20120410#l-358 |
| 16:25 | <MikeSmith> | please don't send me angried DMs |
| 16:25 | <karlcow> | TabAtkins: do you know the bug number on the webkit project for CSS variables? |
| 16:25 | <TabAtkins> | TOO LATE |
| 16:25 | <TabAtkins> | karlcow: I do not. |
| 16:25 | <MikeSmith> | I was referring to http://krijnhoetmer.nl/irc-logs/whatwg/20120410#l-355 |
| 16:26 | <MikeSmith> | whoever's writing self-closing script tags needs to wear a dunce cap and be subjected to a struggle session |
| 16:27 | <karlcow> | hmmm https://bugs.webkit.org/show_bug.cgi?id=46594 |
| 16:27 | <karlcow> | Bug 46594 - [Meta] Finish CSS Variables implementation |
| 16:27 | <karlcow> | this is meta indeeed |
| 16:29 | <MikeSmith> | I like the way Joe Peck puts the quotes around "www-style" there |
| 16:29 | <MikeSmith> | reminds me of the way Bjorn likes to write HTML5 as "HTML5" |
| 16:30 | <TabAtkins> | Heh, that's... not a useful bug. |
| 16:47 | <karlcow> | https://bugs.webkit.org/show_bug.cgi?id=19660 the old version of css variables |
| 16:54 | <karlcow> | ok added css variables https://github.com/karlcow/browserfeatures/blob/master/bugsdb.json#L45 |
| 17:24 | <Hixie> | hsivonen: the even funnier part of that poll is that hober's comment on that post were actually added after the poll closed, because he didn't notice the poll was closing |
| 17:24 | <Hixie> | hsivonen: gotta love the level of engagement the wg is getting |
| 17:42 | <annevk> | btw, is there a smarter way to do width = len(str(len(list)))? |
| 17:42 | <Ms2ger> | Eh |
| 17:42 | <Ms2ger> | The number of digits in the length of the list? |
| 17:43 | <annevk> | yeah, I use it for the width of the first column of the indexes |
| 17:43 | <Ms2ger> | That would be around log10(len(list)) |
| 17:43 | <Ms2ger> | Not sure if it's called that in python |
| 17:44 | <annevk> | it's not really slow |
| 17:44 | <annevk> | was just wondering |
| 17:45 | <TabAtkins> | Yeah, log is what you need if you dont' want string parsing. |
| 17:45 | <TabAtkins> | Specifically, floor(log10(len(list)-1))+1 |
| 17:45 | <Ms2ger> | There's no string *parsing* involved, per se |
| 17:45 | <TabAtkins> | Yeah. |
| 17:46 | <annevk> | so that looks way uglier :) |
| 17:46 | <TabAtkins> | Indeed it does. |
| 17:47 | <Ms2ger> | Are you sure those signs are right? |
| 17:47 | <TabAtkins> | Oh wait, remove one of those fenceposts. |
| 17:48 | <TabAtkins> | floor(log10(len(list)))+1 |
| 17:48 | <TabAtkins> | I was remembering code that was doing that from the last index, rather than the length. |
| 17:48 | <Ms2ger> | list = [] |
| 17:48 | <TabAtkins> | Screw your zero. |
| 18:46 | Philip` | is surprised to discover an OpenGL extension (GL_NV_path_rendering) whose API accepts SVG path strings (and PostScript path strings) |
| 18:48 | <Philip`> | (Seems to be supported on GeForce 8 and higher) |
| 18:49 | Philip` | wonders if it's meant to be used for hardware-accelerating actual SVG drawing |
| 18:51 | <smaug____> | Nvidia has been hiring browser+graphics experts |
| 18:52 | <annevk> | btw |
| 18:52 | <annevk> | the equivalent of my len(str(len(list))) hack is |
| 18:52 | <annevk> | import math |
| 18:52 | <annevk> | math.floor(math.log(len(list), 10))+1 |
| 18:53 | <annevk> | that it's longer is about the nicest thing I can say about it |
| 18:55 | <Philip`> | annevk: If you're trying to measure the width of a column of text, the len(str(...)) approach seems much more semantically appropriate |
| 18:56 | <annevk> | my intuition is usually okay, but then I second guess myself into hell |
| 19:00 | <gsnedders> | smaug____: Any idea to work on what? |
| 19:02 | <smaug____> | dunno. I just happened to notice that they were hiring such devs to their Helsinki office |
| 19:05 | <hsivonen> | Philip`: leaking SVG/PS down to OpenGL seems weird |
| 19:06 | <hsivonen> | Philip`: is it assumed that the app sets the transform matrix so that original path coordinates can go unchanged all the way to GL? |
| 19:20 | <Hixie> | annevk: which is faster? |
| 19:24 | <Philip`> | hsivonen: Paths are constructed in a 2D plane with the specified coordinates, then multiplied by the global 3D modelview-projection-viewport matrices when they are rasterised |
| 19:24 | <Philip`> | hsivonen: so presumably the application is expected to use those matrices |
| 19:25 | <Philip`> | hsivonen: Oh, looks like there's also a TransformPath function which you can use as well as those global matrices |
| 19:26 | Ms2ger | pokes Philip` |
| 19:27 | Philip` | falls off a cliff and dies |
| 19:27 | <annevk> | Hixie: haven't benchmarked |
| 19:27 | <Ms2ger> | We can't have that |
| 19:28 | <Ms2ger> | Philip`, puh-lease have a look at http://www.w3.org/Bugs/Public/show_bug.cgi?id=14191 |
| 19:33 | <Hixie> | hmm |
| 19:33 | <Hixie> | should modal <dialog>s have a UA-implemented "cancel" mechanism, or should it be implemented by the author... |
| 19:33 | <Hixie> | i guess if we add <form method=dialog> then we should have a UA cancel too |
| 19:34 | <annevk> | letting clicking next to dialog dismiss it would sure be nice |
| 19:36 | <karlcow> | β python -mtimeit -s 'x=range(10);from math import ceil,log' 'ceil(log(len(x),10))+1' |
| 19:36 | <karlcow> | 1000000 loops, best of 3: 0.513 usec per loop |
| 19:36 | <karlcow> | β python -mtimeit -s 'x=range(10)' 'len(str(x))' |
| 19:36 | <karlcow> | 100000 loops, best of 3: 2.59 usec per loop |
| 19:37 | <Hixie> | annevk: by default? |
| 19:37 | <Hixie> | annevk: certainly i agree we should have an event to allow that, but i dunno about doing it by default |
| 19:40 | <annevk> | Hixie: dunno, my main interaction with modal dialogs is twitter |
| 19:40 | <annevk> | Hixie: twitter does that :) |
| 19:41 | <Philip`> | Ms2ger: The patch looks like it'll probably break non-W3C versions of the tests (undefined references to assert_true etc) - is that correct and/or intentional? |
| 19:41 | <Ms2ger> | Those exist? |
| 19:41 | <Ms2ger> | Oh |
| 19:41 | <Ms2ger> | Well, I guess I maintain one |
| 19:42 | <Philip`> | Apart from the Mochitest version, I don't think the others are currently used, so it's okay to intentionally break them |
| 19:42 | <Hixie> | annevk: doing it by default seems a bit dodgy for non-trivial dialogs |
| 19:43 | <Ms2ger> | I haven't checked... Let me have a look |
| 19:44 | <Philip`> | (The other versions were used at philip.html5.org but that's obsolete now and not updated) |
| 19:44 | <Ms2ger> | Philip`, Mochitests don't call _assert*, I don't think |
| 19:44 | <Philip`> | (though at least the philip.html5.org versions have a usable test-runner) |
| 19:45 | <Philip`> | Ms2ger: That's quite possibly true |
| 19:46 | <annevk> | Hixie: maybe there should be some way to signal the difference in markup? infobox versus requires user interaction |
| 19:46 | <Hixie> | hmm |
| 19:47 | <annevk> | Hixie: it seems the style features you get are desirable either way |
| 19:47 | <Hixie> | in other news, have you picked the name for the backdrop pseudo? |
| 19:47 | <annevk> | I called it ::backdrop |
| 19:47 | <annevk> | :) |
| 19:47 | Hixie | considers onbackdropclick |
| 19:47 | <annevk> | http://dvcs.w3.org/hg/fullscreen/raw-file/tip/Overview.html#::backdrop-pseudo-element |
| 19:47 | <Ms2ger> | Philip`, testharness.js has at least three test runners, fwiw ;) |
| 19:48 | <annevk> | I guess I should file a bug on HTML for <iframe allowfullscreen> and generally integrating Fullscreen Hixie? |
| 19:48 | <Hixie> | (oh hey maybe you should spec the 'backdropclick' event actually) |
| 19:48 | <Hixie> | lgtm |
| 19:49 | <annevk> | I want the person that defines hit testing and the box model to do things like that... |
| 19:49 | <TabAtkins> | Wait, why backdropclick? |
| 19:49 | <annevk> | but I guess as long as we're patching the broken patchwork that's CSS... |
| 19:49 | <annevk> | I can do that |
| 19:49 | <Ms2ger> | !summon tantek |
| 19:50 | <Hixie> | TabAtkins: so you know when to close the dialog |
| 19:50 | <Hixie> | TabAtkins: if you want to |
| 19:51 | <TabAtkins> | Hm. We're speccing that click events on pseudos dispatch to the superior parent, with a "pseudo" argument on the event object indicating which pseudo was clicked. |
| 19:51 | <TabAtkins> | Presumably the <dialog> is the superior parent of ::backdrop. |
| 19:52 | <Philip`> | Ms2ger: The _assertPixelApprox assertion message is missing the tolerance value, and statements should end in semicolons to be consistent, and maybe it'd be less ugly to put the failure messages in a 'var msg' instead of duplicating the code or something like that, but otherwise I don't see anything obvious enough to notice (though I find it hard to remember how the test system works) |
| 19:52 | <annevk> | TabAtkins: where is that? |
| 19:52 | <annevk> | ooh, so click would dispatch on dialog |
| 19:53 | <annevk> | and it would have .cssPsuedo set to "backdrop"? |
| 19:53 | <TabAtkins> | annevk: Trying to find it. dbaron was talkinga bout it. |
| 19:53 | <TabAtkins> | Yeah. |
| 19:53 | <annevk> | kind of nice |
| 19:53 | <Hixie> | that works for me |
| 19:54 | <annevk> | someone still needs to define hit testing on boxes though |
| 19:54 | <annevk> | along with the mouse events and such |
| 19:54 | Hixie | giggles has his silly term "superior parent" is still in the vernacular |
| 19:54 | <Ms2ger> | Philip`, ok, thanks |
| 19:54 | <annevk> | heh |
| 19:54 | <annevk> | in DOM we'd have called that "associated element" or some such |
| 19:54 | <TabAtkins> | Hixie: I can't come up with a better term. :/ |
| 19:55 | <Hixie> | yeah i couldn't either :-) |
| 19:55 | <TabAtkins> | pseudo-parent! |
| 19:55 | <TabAtkins> | Or pseudo-root. |
| 19:55 | <TabAtkins> | The element that roots the pseudo-tree the element is in. |
| 19:55 | <Ms2ger> | annevk, yeah, we associate a lot :) |
| 19:55 | <annevk> | we keep it simple :) |
| 20:13 | <karlcow> | hmmm can't find http://html5.org/tools/web-apps-tracker?from=7030&to=7031 in http://dev.w3.org/html5/2dcontext/ |
| 20:13 | <karlcow> | I wonder if I'm looking at the wrong place |
| 20:13 | <Ms2ger> | Yep |
| 20:14 | <Ms2ger> | You're looking at <karlcow> hmmm can't find http://html5.org/tools/web-apps-tracker?from=7030&to=7031 in http://dev.w3.org/html5/2dcontext/ |
| 20:14 | <hober> | karlcow: i assume syncing the 2dcontext draft is paused pending resolution of issue-201 |
| 20:14 | <Ms2ger> | Er |
| 20:14 | <Ms2ger> | Ar http://dev.w3.org/html5/2dcontext/ |
| 20:14 | <karlcow> | ah thanks hober |
| 20:15 | Ms2ger | stops trying to type |
| 20:17 | <karlcow> | http://www.w3.org/html/wg/tracker/issues/201 |
| 21:22 | <ilhan> | Hi there! Is there anybody who will be interested to develop the Distributed Social Networking Protocol (dsnp) http://www.complang.org/dsnp/ |
| 21:33 | <karlcow> | ilhan: http://www.complang.org/redmine/projects/dsnpd dead link, also http://www.complang.org/redmine/projects/choice-social |
| 21:33 | <karlcow> | and project seems dead http://www.complang.org/redmine/projects/dsnp/issues?set_filter=1 |
| 21:37 | <ilhan> | karlcow: I've contacted with the author a half months ago, he is developing the project but very very very slowly now |
| 21:52 | <annevk> | 'The βletβ keyword has been unreserved in JavaScriptCore as reserving it caused compatibility problems.' aah |
| 21:52 | <TabAtkins> | wut |
| 21:53 | <TabAtkins> | Do they... do they think that ES will just *not use* 'let'? |
| 21:54 | <annevk> | just quoting from peter.sh |
| 21:57 | <rubys> | tabatkins: not being a reserved word does *not* mean that it won't be used. |
| 21:58 | <rubys> | it just means that var let=1; will still be valid. |
| 21:58 | <TabAtkins> | Ah, that's right. |
| 21:58 | <rubys> | there are extremely few places in the grammar where "token1 token2" is valid, so sequences that have 'let' immediately followed by a token may still be viable. |
| 21:58 | <TabAtkins> | Oh yeah, I remember some of this stuff. ES6 is trying to unreserve more keywords, too, as there's often no reason to prevent a keyword from being an array key or such. |
| 21:59 | <TabAtkins> | object key, that is. |
| 21:59 | <rubys> | understood |
| 21:59 | <rubys> | some languages have no reserve words (REXX is one) |
| 21:59 | <rubys> | reserved* |
| 22:00 | <annevk> | ah yeah, I think that tripped me up actually in ES a number of times |
| 22:00 | <annevk> | and more recently in Web IDL |
| 23:45 | <Hixie> | hey TabAtkins, yt? is there a formal way of saying "apply the table at http://www.w3.org/TR/CSS21/visuren.html#dis-pos-flo " these days? |
| 23:46 | <TabAtkins> | What's the context? You have an input display type, and want the computed value from the table? |
| 23:46 | <Hixie> | yeah |
| 23:46 | <Hixie> | i'm adding a new 'position' keyword like you suggested |
| 23:47 | <TabAtkins> | No formal way beyond what you just quoted. |
| 23:47 | <Hixie> | but need it to convert 'display' to block |
| 23:47 | <Hixie> | k |
| 23:47 | <Hixie> | so it's not in a css3 draft yet or anything |
| 23:47 | <TabAtkins> | Nah. |
| 23:47 | <Hixie> | k |
| 23:47 | <Hixie> | thanks |
| 23:47 | <TabAtkins> | That would be the Box Module, which no one wants to work on yet. |
| 23:47 | <TabAtkins> | Which means I think I'll do it at the end of the year or so. |
| 23:47 | <Hixie> | brave man |
| 23:51 | <Hixie> | ok! |
| 23:51 | <Hixie> | <dialog> is done, i believe |
| 23:51 | <Hixie> | oh crap |
| 23:51 | <Hixie> | i forgot <form method=dialog> again |
| 23:56 | <Hixie> | i wonder if i should somehow make method=dialog invalid if the <form> isn't in a <dialog> |
| 23:56 | <Hixie> | hm, when we talked about this earlier i think we decided it could be useful for other purposes huh |
| 23:57 | <TabAtkins> | Outside of a dialog, it would just be a form that never navigates, right? |
| 23:58 | <Hixie> | yeah |
| 23:59 | <TabAtkins> | Yeah, I'd find that useful. I always forget exactly how I'm supposed to stop navigation, and end up just doing all of them. |
| 23:59 | <Hixie> | fallback behaviour is suboptimal in legacy UAs but not fatally so |
| 23:59 | <Hixie> | yeah i usually end up doing action="javascript:void(0)" |
