| 05:58 | <zcorpan> | hmm, i installed anolis, the testsuite runs fine, but i get "-bash: anolis: command not found" in a new shell |
| 06:55 | <annevk> | hmm |
| 07:54 | <annevk> | is there a shorter way to write: "if either ns is non-null and list is non-empty, or list is non-empty and does not contain ln" |
| 07:55 | <annevk> | "if either ns non-null or list does not contain ln, and list is non-empty" |
| 07:56 | <annevk> | "if list is non-empty and either does not contain ln or ns is non-null" |
| 07:56 | <annevk> | thanks everyone :) |
| 08:02 | <foolip> | you're welcome! |
| 08:02 | <foolip> | annevk, I forgot to CC you on http://lists.w3.org/Archives/Public/public-html-ig-zh/2012Apr/0061.html |
| 08:05 | <annevk> | interesting |
| 08:05 | <annevk> | guess we should just go with hkscs then |
| 08:06 | <annevk> | and maybe not emit the hkscs range from the encoder (using "html entities" instead) |
| 08:06 | <annevk> | although I'm not sure if that's actually worth it |
| 08:07 | <Ms3ger> | zcorpan: did you install anolis or just clone the repo? |
| 08:11 | <foolip> | annevk, just hkscs and the simplest possible encoder seems good to me, if someone can come up with something for uao we can always split the encoders at that point |
| 08:13 | <annevk> | you mean split the decoders? |
| 08:15 | <zcorpan> | Ms3ger: i cloned and then ran sudo python setup.py install |
| 08:15 | <zcorpan> | Ms3ger: btw http://simon.html5.org/test/quirks-mode/reftest/font-element-text-decoration-color/ |
| 08:18 | <jgraham> | zcorpan: Well presumably the script isn't in your path |
| 08:19 | <jgraham> | So I would compare the location of the script to your $PATH |
| 08:22 | <zcorpan> | $ echo $PATH |
| 08:22 | <zcorpan> | /opt/local/bin:/opt/local/sbin:/usr/bin:/bin:/usr/sbin:/sbin:/usr/local/bin:/usr/X11/bin |
| 08:23 | <zcorpan> | changing mode of /opt/local/Library/Frameworks/Python.framework/Versions/2.5/bin/anolis to 755 <-- is this the path i should add? |
| 08:23 | <zcorpan> | minus /anolis |
| 08:25 | <zcorpan> | ok that made anolis throw complaining that lxml: command not found |
| 08:25 | <zcorpan> | so i guess i need to put that in my PATH as well |
| 08:26 | <jgraham> | No.. |
| 08:26 | <jgraham> | I mean lxml shouldn't be a binary |
| 08:28 | <zcorpan> | oh |
| 08:28 | <jgraham> | So, to be honest, my advice would be "if you want to get there, don't start from here" |
| 08:28 | <zcorpan> | sorry, misread the error. ImportError: No module named lxml |
| 08:29 | <jgraham> | Ah, that sounds more possible |
| 08:29 | <jgraham> | I don't know where your lxml is installed or what the library search path for that version of python is |
| 08:30 | <jgraham> | One way to avoid all these problems would be to install everything you need into a virtualenv |
| 08:30 | <jgraham> | Then you would have to activate the virtualenv in whatever whell you use anolis from |
| 08:31 | <jgraham> | But it would all be very tidy |
| 08:34 | <foolip> | annevk, yes if someone suggests a sensible way to split the decoders, we can also split the encoders |
| 08:34 | <hsivonen> | It disappoints me that a suggesting to add heuristic encoding detection to WebVTT even shows up |
| 08:34 | <foolip> | hsivonen, where has that shown up? |
| 08:34 | <hsivonen> | foolip: whatwg list |
| 08:36 | <foolip> | oh, that's weird coming from silvia |
| 08:38 | <annevk> | yeah I don't think we should do that |
| 08:38 | <annevk> | thanks a lot for those notes hsivonen |
| 08:38 | <hsivonen> | annevk: you're welcome |
| 08:39 | <annevk> | i'll study them in a couple of hours, have to go for a bit |
| 08:42 | <hsivonen> | we (Mozilla) keep making questionable decisions when it comes to encodings, because East Asia is treated as a mystery without a proper feedback loop |
| 08:42 | <hsivonen> | i.e. it's assumed that if we break something, we won't get bug reports--just lose market share |
| 08:43 | <annevk> | well for big5 you did act on some feedback, but apparently not on data or without proper check with other browsers |
| 08:44 | <annevk> | yeah that's why sicking didn't want to make HTML parsing in XHR utf-8-only I think |
| 08:44 | <annevk> | but that was not a market share decision, just a being able to use legacy documents or some such |
| 08:45 | <smaug____> | hsivonen: really ? |
| 08:45 | <smaug____> | we do have an office in Japan where there are active core developers |
| 08:46 | <foolip> | hsivonen, indeed, if Firefox hadn't bought into the terrible idea of Big5-UAO it would certainly have been dead and forgotten :( |
| 08:47 | <hsivonen> | smaug____: yes, really. :-( |
| 08:50 | <smaug____> | hsivonen: I'd like to know an example |
| 08:50 | <hsivonen> | annevk: well, the decisions to get rid of the 1024 byte limit for the detector was base on a single report (Japanese Planet Debian) plus second or third hand hearsay instead of someone from the Japan office presenting solid data |
| 08:51 | <hsivonen> | smaug____: 1) sicking's arguments for not falling back to UTF-8 in XHR. 2) the panic treatment of the heuristic detection limit in the Firefox 4 cycle. 3) perhaps the big5 stuff |
| 08:54 | <zcorpan> | hsivonen: is it possible to grep for problematic pages (for the 1024 limit)? |
| 08:54 | <hsivonen> | zcorpan: I doubt grep works |
| 08:54 | <hsivonen> | zcorpan: it would probably need a more complicated harness that actually runs chardet |
| 08:55 | <hsivonen> | smaug____: also potentially the FileReader stuff |
| 08:55 | <zcorpan> | yeah, k |
| 08:55 | <hsivonen> | I don't know what the rationale for the FileReader thing is |
| 08:56 | <hsivonen> | smaug____: also the BOMless UTF-16 Basic Latin thing was a panic from a single data point. (Albeit from India rather than East Asia.) |
| 08:57 | <hsivonen> | smaug____: we get a single weird data point from Asia and rush in a browser side fix. We get a single weird data point from Hungary or Romania and we'll shrug it as something to take a look at later |
| 08:58 | <hsivonen> | *shrug it off |
| 08:58 | <smaug____> | hsivonen: well if you think we should do something better in this area, I hope you've filed a meta bug to fix all these issues properly ;) |
| 08:58 | <smaug____> | and raised the issue in newsgroups |
| 08:59 | <hsivonen> | last time I tried to inject some data-driven decision making in the process, I was told I can't obtain the data due to privacy |
| 09:02 | <hsivonen> | also, when the sv-SE localization had bad default, I was able to say that things aren't supposed to be that weird in Sweden |
| 09:03 | <hsivonen> | but for random other localization it's assumed that the localizer knows better than I |
| 09:36 | <zcorpan> | btw, seems the font element text decoration quirk is the first quirk i'm writing tests for that one browser completely passes (webkit passes) |
| 09:37 | <zcorpan> | oh, no it wasn't (blocks ignore line height passes in opera/webkit/gecko) |
| 09:38 | <AryehGregor> | hsivonen, what was the data you wanted to obtain but couldn't due to privacy? |
| 09:41 | <hsivonen> | AryehGregor: Changes to fallback encoding on a per-locale basis as part of telemetry |
| 09:41 | <AryehGregor> | How would that violate privacy? This is opt-in telemetry only, right? |
| 09:43 | <hsivonen> | AryehGregor: https://groups.google.com/d/msg/mozilla.dev.platform/CFDMuKOLyBE/xhaE0_e0JlEJ |
| 09:43 | <hsivonen> | AryehGregor: the telemetry opt-in isn't opt-in to whatever |
| 10:04 | <hsivonen> | the new "Browser Version (partially combined)" option doesn't seem combined to me on statcounter |
| 10:11 | <annevk> | okay, so besides the "universal" detector you can also enable a subset of them and which is depends on the local of the end user |
| 10:11 | <annevk> | locale* |
| 10:19 | <annevk> | foolip: I don't get where the FFED in your big5 script is coming from |
| 10:55 | <kennyluck> | typo of FFFD ? |
| 10:57 | <annevk> | found it foolip |
| 10:57 | <annevk> | foolip: didn't realize you had made some other changes to the handling of the data besides manually adding a few code points |
| 11:02 | <annevk> | kennyluck: see https://www.w3.org/Bugs/Public/show_bug.cgi?id=16822 |
| 11:08 | <kennyluck> | annevk, thanks |
| 11:20 | <hsivonen> | hmm. still no reply to my email to coremob asking about the purpose/meaning of ring 0 |
| 11:26 | <annevk> | that <var> and <em> have the same styling can be a little confusing sometimes |
| 11:26 | <annevk> | maybe we should style <var> in standards with something extra |
| 11:27 | <hsivonen> | Is Opera expected to say "no style information" when as XSLT transform is bogus? http://hsivonen.iki.fi/test/moz/readystate/xslt-fail.html |
| 11:29 | <annevk> | I'm not sure we have XSLT error reporting |
| 11:35 | <jgraham> | hsivonen: I think so, but I don't really know why] |
| 11:36 | <hsivonen> | ok |
| 11:36 | <jgraham> | I could ask but the person mos likely to know just went for lunch |
| 11:36 | <jgraham> | Which is a good idea, all things considered |
| 11:37 | <hsivonen> | good idea to get food or good idea to dodge XSLT-related questions? |
| 11:37 | <jgraham> | Yes |
| 11:38 | <jgraham> | Both :) |
| 11:40 | <hsivonen> | I wish there was an easy way to query hg for "show all .xsl files I've ever added" |
| 11:46 | <annevk> | kennyluck: foolip: my idea of the encoder is to determine lead/trail and if lead is less than 0xA1 emit an encoder error |
| 11:50 | <zcorpan> | annevk: em { text-decoration:underline } |
| 11:51 | <hsivonen> | jgraham: the order of events is weird in Opera for http://hsivonen.iki.fi/test/moz/readystate/xslt.html |
| 11:51 | <hsivonen> | jgraham: I thought Opera had the same architecture for XSLT as Trident and WebKit. Apparently not quite so. |
| 11:52 | <hsivonen> | the Gecko results are sad, too |
| 11:52 | <zcorpan> | annevk: or var { font-family:serif } |
| 11:52 | <hsivonen> | but it's less surprising considering Gecko's different XSLT architecture |
| 11:55 | <hsivonen> | so I'm thinking of forming an opinion about readyState |
| 11:55 | <hsivonen> | * Aborted docs should reach "complete" |
| 11:56 | <hsivonen> | * XSLT errors should reach "complete" |
| 11:56 | <hsivonen> | * Aborted docs should not fire "load" |
| 11:57 | <hsivonen> | * XSLT errors should fire "load" |
| 11:57 | <hsivonen> | * It shouldn't be possible to transition from "loading" to "complete" without going through "interactive" |
| 11:57 | <hsivonen> | * All docs should fire "DOMContentLoaded" |
| 11:58 | <hsivonen> | * "load" should never fire synchronously |
| 11:58 | <zcorpan> | or font-family:cursive for more fancyness |
| 11:58 | <hsivonen> | * It's OK for the transition to "interactive" to be synchronous when document.close() processes the EOF synchronously |
| 11:59 | <hsivonen> | * window.stop() should abort |
| 11:59 | <hsivonen> | do the above points make sense? worth posting to the list? |
| 12:04 | <smaug____> | hsivonen: could you perhaps indicate which browsers do what |
| 12:04 | <annevk> | zcorpan: cannot underline things other than links I think and serif is ugly, but thanks for the suggestions :) |
| 12:07 | <zcorpan> | come on, cursive looks awesome :-) |
| 12:07 | <hsivonen> | smaug____: ok. I'll put rationale in my post |
| 12:09 | <annevk> | font-family:cursive ? what is that |
| 12:09 | <annevk> | hmm |
| 12:12 | <smaug____> | hsivonen: could you pastebin the results even now? |
| 12:12 | <smaug____> | or put them to some wiki |
| 12:13 | <hsivonen> | smaug____: the results are not in a pasteable format |
| 12:14 | <hsivonen> | smaug____: that is, I'd have to run all the demos again |
| 12:14 | <smaug____> | ok |
| 12:14 | smaug____ | waits for the email |
| 12:15 | <smaug____> | though, if the data is emailable format, it should be pastebinable too :) |
| 12:18 | <smaug____> | hmm, if I have a document fragment, what is the easiest way to serialize it |
| 12:18 | <smaug____> | its content, I guess |
| 12:18 | <hsivonen> | smaug____: I wasn't going to email the test output |
| 12:19 | <smaug____> | hsivonen: it would be helpful to know what browsers do now |
| 12:20 | <smaug____> | s/helpful/useful/ |
| 12:25 | <annevk> | kennyluck: foolip: http://dvcs.w3.org/hg/encoding/raw-file/tip/Overview.html#big5-encoder |
| 12:28 | <jgraham> | hsivonen: That doesn't obviously look bad to me, but the existing spec doesn't obviously look bad |
| 12:29 | <jgraham> | So it would be helpful if you state the reasoning + the delta from the existing spec |
| 12:29 | <jgraham> | In your email |
| 12:40 | <kennyluck> | annevk, I like the fact that the algorithm avoid emitting hkscs extension. I am not sure how foolip feels about this though. |
| 12:40 | <zcorpan> | annevk: "The EOF byte is a conceptual byte representing the last byte in a stream." - doesn't it represent the end of the stream (and the previous byte is the last byte)? |
| 12:40 | <kennyluck> | annevk, How many CP950 characters are in the < 0xA1range? |
| 12:41 | kennyluck | is pretty ignorant about big5, in fact. |
| 12:48 | <annevk> | kennyluck: zero afaik |
| 12:48 | <annevk> | zcorpan: fair point |
| 13:05 | <kennyluck> | hmm, some of the many to one mappings are removed in big5. I think that's a good idea, but given that I don't understand why CP950 did this at the beginning, there's nothing much I could say. |
| 13:05 | <annevk> | you mean encoder-wise? |
| 13:06 | <kennyluck> | annevk, yeah. |
| 13:06 | <annevk> | it could be that we need to add that I suppose |
| 13:10 | <kennyluck> | annevk, just as an example. In http://moztw.org/docs/big5/table/cp950-u2b.txt , all 財, ㊖ and ㈶ would be encoded in the sequence that would be decoded as 財. |
| 13:10 | <kennyluck> | (The 0xb0 0x5d sequence) |
| 13:13 | <annevk> | kennyluck: yeah I know browsers do that |
| 13:14 | <annevk> | kennyluck: I have some simple scripts where with some tweaks I can find out what each Unicode code point maps to |
| 13:14 | <annevk> | kennyluck: takes forever to run though if I go beyond BMP |
| 13:14 | <annevk> | I should work on publishing those scripts somehow |
| 13:16 | <kennyluck> | annevk, anyway, I don't think the current algorithm is inferior in theory, so I suggest we expand the description of index-big5.txt so that these differences from CP950 are noted. |
| 13:17 | <kennyluck> | This should include the fact that the index-big5.txt has no PUA mappings too. |
| 13:19 | <annevk> | kennyluck: 1) index big5 is currently not described as being equivalent to cp950 2) cp950 officially does not include PUA |
| 13:20 | <annevk> | kennyluck: not sure if cp950 officially includes fallback mappings |
| 13:26 | <kennyluck> | annevk, well, people will want to compare the algorithm with what's implemented. That includes the PUA and I have no better name for it. Also, the current description says "This matches the Big5 standard in combination with the Hong Kong Supplementary Character Set and other common extensions." but I have no idea what this "Big5 standard" refers to. |
| 13:27 | <annevk> | yeah I should add some references at some point |
| 13:27 | <annevk> | but in practice I suspect people will have to compare it to the internal tables anyway |
| 13:27 | <kennyluck> | nobody knows what big5 is :( |
| 13:27 | <annevk> | or via a test suite or some such |
| 13:37 | <zewt> | annevk: any way to run the scripts in workers to speed them up? |
| 13:37 | <zewt> | (don't know what APIs you're using) |
| 13:38 | <annevk> | using URLs |
| 13:39 | <annevk> | and you need document encoding |
| 13:39 | <annevk> | if browsers expose the encoding API at some point and that's guaranteed to use the same mechanism it might work |
| 15:30 | <hsivonen> | got the readyState proposal done: http://lists.whatwg.org/pipermail/whatwg-whatwg.org/2012-April/035521.html |
| 15:31 | <JVoracek> | |
| 15:33 | <jgraham> | hsivonen: Thanks |
| 15:35 | <annevk> | I wonder if this is the only observable attribute from the outside; otherwise we're going to have to do this again and again... |
| 15:39 | <jgraham> | Only observable attribute of what? |
| 15:39 | <annevk> | the various processes going on |
| 16:11 | <dglazkov> | good morning, Whatwg! |
| 16:38 | <gsnedders> | gsnedders is trying do maths, part [lost count years ago]: what's the simplest way to find the sign of a permutation cycle? |
| 16:40 | <annevk> | gsnedders: Google suggests http://www.math.uconn.edu/~kconrad/blurbs/grouptheory/sign.pdf |
| 16:42 | <gsnedders> | "In words, if a cycle has even length then its sign is -1, [otherwise] 1." Why do more places not just state that? |
| 16:42 | <Ms2ger> | Too trivial :) |
| 16:43 | <Hixie> | permutation cycles have signs? |
| 16:43 | <Ms2ger> | odd/even, I think we called them |
| 16:44 | <gsnedders> | Hixie: Yes. http://en.wikipedia.org/wiki/Even_and_odd_permutations |
| 16:45 | gsnedders | is totally looking forward to his exam tomorrow morning |
| 16:45 | <annevk> | I think I got scammed for the first time ever, by some US government-like looking site trying to get me to enter ESTA info |
| 16:45 | <annevk> | ugh |
| 16:45 | <Ms2ger> | You lose one internet, sir |
| 16:45 | <annevk> | I didn't go all the way |
| 16:45 | <TabAtkins_> | You pulled out before the end? |
| 16:45 | <TabAtkins_> | I hear that's not reliable. |
| 16:46 | <annevk> | it's what jesus would do though |
| 16:46 | <TabAtkins_> | Catholic Jesus, at least. |
| 16:46 | <annevk> | fair point |
| 16:46 | <Hixie> | was the site encrypted? |
| 16:46 | <annevk> | yeah |
| 16:46 | <Hixie> | who owned the cert? |
| 16:46 | <Hixie> | maybe they can be reported |
| 16:47 | <Hixie> | to their authority |
| 16:47 | <Ms2ger> | Anyone who feels like doing a little WebIDL in their head? |
| 16:47 | <Ms2ger> | var list = document.body.children |
| 16:47 | <Ms2ger> | Object.defineProperty(list, "length", {get: function() { throw 7 }}) |
| 16:47 | <Ms2ger> | new Blob(list) |
| 16:48 | gsnedders | slaps Ms2ger for even writing that |
| 16:48 | <annevk> | Hixie: https://www.nationalvisas.com/ RapidSSL CA; might actually be a legal thing I suppose, they just charge more than usual |
| 16:48 | <Ms2ger> | gsnedders, thank you :) |
| 16:49 | <Ms2ger> | That proves I'm doing it right :) |
| 16:49 | <gsnedders> | Ms2ger: Masochist. |
| 16:49 | <Ms2ger> | I wonder, will Opera get it right? :) |
| 16:50 | <Hixie> | annevk: yeah, dunno |
| 16:50 | <Hixie> | annevk: might be "legit" |
| 16:50 | <Hixie> | for some definition thereof |
| 16:53 | <annevk> | whoa |
| 16:53 | <annevk> | certificate dialogs are application global in Chrome |
| 16:53 | <annevk> | (once you click on certificate information) |
| 17:01 | <Hixie> | good lord, i just hit inbox-unread-zero for the first time in weeks |
| 17:05 | <annevk> | whoa, that's bad |
| 17:06 | <annevk> | I've been dropping mailing lists and it seems to be working out |
| 17:07 | <TabAtkins_> | I've been doing the opposite. I still hit unread-0 every morning, but it takes a while (especially on Mondays), and I have to make concerted efforts to burn down the read-but-not-acted-on pile. :/ |
| 17:07 | <TabAtkins_> | But I felt a compulsion to start interacting with ES, so shrug. |
| 17:07 | <gsnedders> | TabAtkins_: es-discuss is heavy going. |
| 17:07 | <annevk> | ooh es-discuss |
| 17:07 | <TabAtkins_> | Not as bad as I would have thought. |
| 17:07 | <annevk> | and www-style |
| 17:07 | gsnedders | has all but given up on almost all mailing lists :P |
| 17:07 | <annevk> | you're crazy |
| 17:08 | <TabAtkins_> | The ES chairs are very reasonable. |
| 17:08 | <TabAtkins_> | Allen and Brendan. |
| 17:08 | <gsnedders> | That doesn't in itself make it easy to keep up with. |
| 17:09 | <TabAtkins_> | True, but that's what strategic use of muting is for. |
| 17:09 | <gsnedders> | I guess my issue is more just a lot of it requires more thought than www-style does, for example. |
| 17:10 | <TabAtkins_> | True. |
| 17:10 | <annevk> | what's the status of http://norbertlindenberg.com/2012/03/ecmascript-supplementary-characters/index.html ? |
| 17:10 | <TabAtkins_> | The arguments over the weekend over soft-binding or hard-binding "this" for fat-arrow functions, for example, were quite subtle. |
| 17:10 | <annevk> | JavaScript guys changing the string representation of the entire platform within the TC39 bubble is somewhat scary / annoying |
| 17:11 | <gsnedders> | annevk: The only two places it matters are at an ES level and a WebIDL level one, no? And it's been discussed on mailing lists for both. |
| 17:12 | <annevk> | well it matters e.g. for <textarea maxlength> |
| 17:12 | <gsnedders> | annevk: Surely that isn't defined in terms of the ES representation of a string, though? |
| 17:13 | <annevk> | it's defined in code units I think |
| 17:13 | <gsnedders> | So you hit maxlength with a string of a shorter length. |
| 17:13 | <annevk> | depends on how you define string |
| 17:13 | <annevk> | but in the platform string means 16bit code units |
| 17:13 | <gsnedders> | the primitive String type. |
| 17:14 | <annevk> | because of ES /browser architecture, etc. |
| 17:16 | <gsnedders> | http://www.w3.org/mid/4F40B3ED.5020604⊙mc is outwith the TC39 bubble. |
| 17:18 | <annevk> | yeah I saw that |
| 17:18 | <annevk> | but that's a month ago |
| 17:18 | <annevk> | two months even |
| 17:19 | <annevk> | whereas what I linked above is only a month old |
| 17:19 | <annevk> | with a long discussion on es-discuss |
| 17:19 | <gsnedders> | What discussion on es-discuss? |
| 17:19 | <gsnedders> | Am I just blind? |
| 17:20 | <gsnedders> | Oh, there. |
| 17:20 | <annevk> | https://mail.mozilla.org/pipermail/es-discuss/2012-March/thread.html#21247 |
| 17:20 | gsnedders | notes that complaining about TC39 not doing more cross-WG communication here doesn't help anything |
| 17:29 | <MikeSmith> | annevk: gsnedders is right, you should complain on the #tc39 channel where they all hang out |
| 17:30 | <annevk> | MikeSmith: server? |
| 17:30 | <annevk> | oh wait, sarcasm? |
| 17:30 | <MikeSmith> | heh |
| 17:30 | <MikeSmith> | yeah |
| 17:32 | <annevk> | so that article concludes with |
| 17:32 | <annevk> | "To really help developers, the focus shouldn’t be on access to individual code points. It should be on more and better functions to process text at higher levels of abstractions. Regular expressions with support for Unicode properties and grapheme clusters would be an excellent start." |
| 17:32 | <annevk> | which I agree with |
| 17:32 | <annevk> | and it also doesn't do the weird global flip people were considering |
| 17:33 | <annevk> | but if that's the end goal, why bother with code point based string iterator and some codePointAt methods... |
| 17:34 | <gsnedders> | http://www.w3.org/mid/CAP40CR28yADVHQgALA3numziGKy_KWSKBjetZ5YG=zSavPyuAQ⊙mgc I basically agree with, FWIW |
| 17:39 | <jsbell> | If ES6 implemented that proposal my encoding impl shim would get to drop a whopping 6 or so lines of code. |
| 17:40 | <annevk> | worth it :p |
| 17:41 | <annevk> | but yeah, once you figure it out the code unit -> code point math is not that hard |
| 17:41 | <annevk> | a problem might be that not many people are doing it, but they're also not considering grapheme clusters so how bad can it be |
| 17:41 | <jsbell> | Snark aside, I'm a fan of the proposal. IMHO a little more formalization of ES strings as UTF-16 is worth it. |
| 17:43 | <annevk> | I would kind of hate it if at some point we have length exposed as code units in one place and code points in another |
| 17:43 | <annevk> | and I can totally see that happening as people typically think they're smarter than whatever went before |
| 17:44 | <annevk> | but this proposal does seem reasonable |
| 17:48 | <annevk> | jsbell: "csueckr" is the wrong spelling, the spec changed it to "cseuckr" |
| 17:49 | <annevk> | jsbell: unfortunately that also changed order |
| 17:54 | <jsbell> | annevk: thanks, fixed. I'll refactor soon to use the encodings.json/indexes.json almost directly. |
| 17:57 | <annevk> | ah cool, that'll fix your big5 index bug I fixed this morning |
| 19:01 | <krit> | Philip`: ping |
| 19:01 | <Philip`> | krit: Pong |
| 19:03 | <annevk> | yay WebKit (for once?) |
| 19:03 | <annevk> | thanks ap, maybe we should strive for something close to what you guys have instead then |
| 19:04 | <krit> | hi Philip`. I looked at your canvas test suite. |
| 19:04 | <ap> | annevk: would be great to eliminate guesswork from decoding |
| 19:04 | <krit> | Philip`: how do you maintain the references in your tests? |
| 19:04 | <krit> | Philip`: As example http://philip.html5.org/tests/canvas/suite/tests/2d.transformation.transform.skewed.html |
| 19:04 | <annevk> | ap: yeah, that's my goal |
| 19:04 | <krit> | Philip`: has a reference to http://philip.html5.org/tests/canvas/suite/tests/spec.html#testrefs.2d.transformation.transform |
| 19:05 | <krit> | Philip`: Do you add the references manually? |
| 19:05 | <annevk> | removing guesswork, one platform layer at a time :) |
| 19:06 | <Philip`> | krit: (It's probably better to use the version at http://test.w3.org/html/tests/submission/PhilipTaylor/canvas/ (or http://dvcs.w3.org/hg/html/file/tip/tests/submission/PhilipTaylor) since that's somewhat more maintained) |
| 19:06 | <krit> | Phae: the question is more how you maintained this references :) |
| 19:07 | <krit> | Philip`: --^ |
| 19:07 | <Philip`> | krit: There's a manually-written file http://dvcs.w3.org/hg/html/file/tip/tests/submission/PhilipTaylor/tools/canvas/spec.yaml that refers to spec sentences |
| 19:07 | <Philip`> | krit: and then a script that does the matching and spec-annotating |
| 19:08 | <krit> | Philip`: means you have to update this file if the spec text changes? |
| 19:08 | <Ms2ger> | It's all regices |
| 19:08 | <Philip`> | krit: Yes, if it changes enough to break the regexy patterns |
| 19:08 | <Philip`> | (which usually means the tests have to be updated too) |
| 19:09 | <krit> | Philip`: thanks |
| 19:10 | <Philip`> | krit: Back when I bothered maintaining the tests, updating the references didn't take much work, and it helped highlight which tests relied on parts of the spec that had changed |
| 19:10 | <Philip`> | krit: so it's kind of evil but I think it works alright :-) |
| 19:11 | <krit> | Philip`: I can imagine that it is more problematic on a spec like HTML that changes frequently |
| 19:11 | <krit> | Philip`: How much time does it take nowadays to maintain this list? |
| 19:12 | <Philip`> | krit: This *is* [a part of] the HTML spec |
| 19:12 | <Philip`> | krit: The spec as a whole changes frequently, but the individual parts of it change pretty infrequently and usually in large batches |
| 19:13 | <krit> | Philip`: But it is still you who updated the file? |
| 19:13 | <Philip`> | krit: Ms2ger would probably be a better person to ask about what happens nowadays |
| 19:14 | <annevk> | from not ever changing we now have "IE is likely not to change" |
| 19:14 | <krit> | Ms2ger: oh, so your comment was related to the discussion? |
| 19:14 | <Ms2ger> | :) |
| 19:14 | <Ms2ger> | I've been maintaining that code more than I'd like |
| 19:14 | <krit> | Ms2ger: not sure what "regices" means :P |
| 19:14 | <Ms2ger> | It's the plural of "regex", obviously ;) |
| 19:15 | <krit> | ahhh hehe |
| 19:15 | <krit> | Ms2ger: can we talk more later? Have a meeting now (lunch :D) |
| 19:15 | <Ms2ger> | Sure, but not today :) |
| 19:15 | <krit> | np |
| 19:15 | <krit> | Ms2ger: Philip`Thanks for your time |
| 19:16 | <Ms2ger> | Np |
| 20:17 | <sftrabbit> | Is it a known bug that the w3c validator doesn't seem to check the microformats wiki for valid rel link types, even though it links to it as an explanation for the error? |
| 20:18 | <MikeSmith> | sftrabbit: it's intentional |
| 20:18 | <sftrabbit> | why's that? |
| 20:18 | <MikeSmith> | we update the validator based on changes to the wiki |
| 20:19 | <MikeSmith> | because it's more practical |
| 20:19 | <MikeSmith> | if you have a specific rel value that we're not supporting, please file a bug |
| 20:19 | <MikeSmith> | or let me know and I'll take a look at it right now |
| 20:22 | <sftrabbit> | "logo" and "sitemap" are flagging up for me |
| 20:22 | <sftrabbit> | on link elements |
| 20:29 | <annevk> | whoa Chrome <input type=date> has terrible UI if you start editing the field |
| 20:29 | <annevk> | by keyboard |
| 20:29 | <annevk> | why even make it editable if pretty much any change wipes it? |
| 20:29 | <smaug____> | type=date is hard |
| 20:30 | smaug____ | waits to see the first good implementation |
| 20:30 | <annevk> | we've had it since 2006 |
| 20:30 | <smaug____> | since 2006 doesn't mean it is good ;) |
| 20:31 | <annevk> | yeah well, nothing is perfect |
| 20:32 | <annevk> | but at least with Opera's implementation you can only enter a date and not confuse it for an arbitrary field |
| 20:32 | <annevk> | whereas in Chrome's implementation you likely lose your textual input until you notice that arrow |
| 20:33 | <annevk> | and if you then manually modify it you might still be out of luck if you don't really know American date conventions |
| 21:19 | <jgraham> | Oh, drama |
| 21:19 | <jgraham> | Guess that means it's time for sleep |
| 21:25 | <othermaciej> | drama? where? |
| 21:47 | <kennyluck> | WHATCG |
| 21:48 | <othermaciej> | clearly the next step is to create a Task Force |
| 21:54 | <hober> | WHATWGCGTF |
| 21:55 | <kennyluck> | Can someone clarify that http://lists.w3.org/Archives/Public/public-whatwg/ is not the mailing list we are going to use? |
| 21:56 | <Hixie> | correct |
| 21:56 | <hober> | kennyluck: http://www.w3.org/community/about/faq/#can-a-community-group-use-its-own-infrastructure-not-hosted-by-w3c |
| 21:56 | <Hixie> | whatwg⊙wo is the mailing list |
| 21:56 | <Hixie> | basically nothing changes until we do an FSA publication |
| 21:56 | <Hixie> | and then we can get people to sign the patent policy |
| 21:56 | <Hixie> | if they want to |
| 21:56 | <TabAtkins_> | And I assume you'll kick off FSA publications regularly? |
| 21:56 | <Hixie> | yeah |
| 21:57 | <Hixie> | how often depends on how long it takes for key companies (esp. opera, mozilla, apple, google) to sign the first one |
| 21:57 | <kennyluck> | FWIW, no one mentioned this in these mails and I expect newcomers get confused because the list is closer to the homepage of the CG. |
| 21:57 | <Hixie> | i plan to fix that |
| 21:57 | <Hixie> | as soon as i'm a member of the CG :-P |
| 21:57 | <kennyluck> | :p |
| 21:58 | <hober> | kennyluck: CGs get auto-created once there are N supporters; the autogeneration script doesn't know how to handle cases where the CG already has infrastructure elsewhere. the links will get fixed. |
| 21:58 | Hixie | has to wait for his AC rep to join his company before he can join the CG :-) |
| 21:58 | hober | is in the same boat |
| 21:58 | <TabAtkins_> | At least I'll be able to join the CG once you do without bugging TV. |
| 21:58 | <Hixie> | hober: do you know if we have to get manual intervention to fix those links? |
| 21:59 | <hober> | Hixie: i don't, but I bet MikeSmith does. |
| 21:59 | <Hixie> | k |
| 21:59 | <MikeSmith> | which links? |
| 21:59 | <hober> | MikeSmith: the links from https://www.w3.org/community/whatwg/ |
| 21:59 | <hober> | to the ml etc |
| 22:00 | MikeSmith | takes a look |
| 22:00 | <hober> | those should all point to the whatwg.org equivalents |
| 22:00 | <hober> | and not the autogenerated thigns |
| 22:00 | <hober> | which should get removed |
| 22:00 | <Hixie> | ideally the wiki too |
| 22:00 | <hober> | and chat |
| 22:01 | <Hixie> | adn the irc link yeah |
| 22:01 | <Hixie> | maybe even the blog :-P |
| 22:01 | <Hixie> | but i dunno how that'd work |
| 22:01 | <Hixie> | anyway, no rush |
| 22:45 | <kennyluck> | Does anyone know why the WHATCG can go straightly from a proposed CG to an accepted one? |
| 22:47 | <TabAtkins_> | If you get five signups it's automatically accepted. |
| 22:48 | <kennyluck> | TabAtkins_, I only see two in the list though. |
| 22:48 | <TabAtkins_> | Shrug. |
| 22:50 | <kennyluck> | I am really happy about this news. There's no longer a need to explain the relationship between W3C and WHATWG. Just join the W3C WHATCG! |
| 22:50 | <othermaciej> | it's accepted now |
| 22:51 | <othermaciej> | b/c there were 5 supporters |
| 22:51 | <othermaciej> | but supporting creation does not join you automatically |
| 22:51 | <othermaciej> | and for people who belong to member orgs, they need their AC rep to approve |
| 22:52 | <kennyluck> | I see. It's still a bit mysterious to me that it gets accepted in 30 mins. |
| 23:01 | <Hixie> | kennyluck: it's automated |
| 23:01 | <Hixie> | kennyluck: as far as i can tell |
| 23:01 | <Hixie> | kennyluck: that's what the documentation says anyway |
| 23:01 | <Hixie> | kennyluck: you just need five supporters |
| 23:05 | <zewt> | now if only every cg didn't feel the strange need to set up their own, isolated mailing list ... |