| 00:11 | <GPHemsley> | Speaking of F2Fs... is anyone going to IETF 85 in Atlanta in early November? |
| 01:05 | <Yuhong> | <zewt> people sure seem to confuse "the actions of an employee reflect on the company" with "everything an employee says is the opinion of the company" |
| 01:05 | <Yuhong> | <zewt> really doesn't seem like that fine a distinction |
| 01:05 | <Yuhong> | I think it is a relic of legacy PR based on control of the message3. |
| 01:05 | <Yuhong> | I think it is a relic of legacy PR based on control of the message. |
| 01:10 | <SamB_MacG5> | if I want to link to a snapshot of entities.json, how should I do that? |
| 01:11 | <SamB_MacG5> | (preferably something a bit more human-friendly than https://github.com/w3c/html/raw/4b354c25cdc7025fef9f561bbc98fee2d9d241c1/entities.json ?) |
| 01:11 | <TabAtkins> | Unless it's branched, that's the correct way to link to it. |
| 01:11 | <TabAtkins> | You mean a *specific* snapshot, right? |
| 01:12 | <SamB_MacG5> | well, something recent would be good |
| 01:12 | <SamB_MacG5> | but mostly I just don't want to link to something that will change ... |
| 01:13 | <TabAtkins> | Okay, then yeah, you can't get away from that kind of url, unless Robin's done a tag branch (which you can then refer to the tip of). |
| 01:13 | <SamB_MacG5> | tag *branch*? |
| 01:14 | <TabAtkins> | I may be misusing terms. |
| 01:14 | SamB_MacG5 | was afraid someone was misusing git |
| 01:52 | SamB_MacG5 | wonders how to do a strut in HTML |
| 04:27 | <divya> | why is this TPAC registration form from 1996? |
| 04:27 | <divya> | the url actually says 2002 |
| 04:27 | <divya> | but NEVER MIND THAT. |
| 04:28 | <SamB_MacG5> | maybe they use the same form every year ? |
| 04:28 | <SamB_MacG5> | or <insert interval of repetition here> |
| 04:29 | <divya> | but it is so unfriendly and unusable :| |
| 04:29 | <divya> | success message is in red. |
| 04:29 | <divya> | radio button + checkbox in the same field |
| 04:29 | <divya> | multiple questions to complete. |
| 04:29 | <SamB_MacG5> | oh, this is an online form |
| 04:30 | <divya> | ahahah there are *offline forms*?!?! |
| 04:30 | SamB_MacG5 | has no idea |
| 04:30 | SamB_MacG5 | doesn't even know what TPAC is |
| 04:30 | <divya> | SamB_MacG5: tagging is a way to represent a state |
| 04:30 | <divya> | so you dont have to refer to a SHA |
| 04:30 | <divya> | but just refer to that tag |
| 04:30 | <divya> | so you *tag* that state and refer to entities.json with that tag. |
| 04:31 | <divya> | that is unhanging |
| 04:31 | <divya> | unchanging even |
| 04:31 | <SamB_MacG5> | yeah, I know how git tags work ;-) |
| 04:31 | SamB_MacG5 | was just confused by the phrase "tag branch" |
| 04:34 | <divya> | srry SamB_MacG5 i have muddled my way through the logs as usual |
| 04:34 | <SamB_MacG5> | no problem |
| 04:35 | <SamB_MacG5> | just didn't want you to waste any more time explaining it to me ;-) |
| 05:35 | <annevk> | and suddenly there is a lot of email |
| 05:35 | <annevk> | GPHemsley: was for wiki cleanup |
| 06:14 | <annevk> | arun really shouldn't resolve those bugs as fixed... |
| 06:39 | <annevk> | Hixie_: maybe make http://spec.whatwg.org/ redirect to http://www.whatwg.org/specs/ ? |
| 07:01 | <annevk> | https://twitter.com/robinberjon/status/255307260375494656 haha why did I miss this |
| 07:12 | <hsivonen> | yay hardware debugging https://bugzilla.mozilla.org/show_bug.cgi?id=800533#c15 |
| 07:24 | <MikeSmith> | hsivonen: excellent |
| 07:25 | <MikeSmith> | somebody should post that to https://twitter.com/BestOfBugzilla |
| 07:42 | <othermaciej> | hsivonen: the interesting thing is that this turned out to be relevant to the cause of the bug but was initially ignored |
| 07:42 | <othermaciej> | (or so it seems from just the bugzilla comments, anyway) |
| 07:43 | <othermaciej> | annevk: that is awesome |
| 07:45 | <hsivonen> | one small step towards encoding sanity: https://hg.mozilla.org/integration/mozilla-inbound/rev/6db369a15ea7 |
| 07:45 | <hsivonen> | boo UTF-16 |
| 07:46 | <annevk> | hsivonen: how could it ever use utf-7? |
| 07:47 | <hsivonen> | annevk: it shouldn’t |
| 07:47 | <annevk> | shouldn't that be removed from IsAsciiCompatible then? |
| 07:48 | <hsivonen> | annevk: It’s basically an in-depth defense as long as the UTF-7 stuff exists in the charset alias system |
| 07:48 | <annevk> | okay |
| 07:49 | <hsivonen> | it scares me that x-imap4-modified-utf7 exists in the alias system |
| 07:49 | <hsivonen> | that sort of thing should not exist outside the IMAP protocol impl, IMO |
| 07:53 | <zcorpan> | MikeSmith: did you see my request about quirks mode tests on w3c-test.org? |
| 07:53 | <MikeSmith> | zcorpan: nope I didn't yet |
| 07:54 | <MikeSmith> | what I need to do? |
| 07:54 | <zcorpan> | MikeSmith: i have tests here i'd like synced on w3c-test.org http://dvcs.w3.org/hg/quirks-mode/file/3397c5f9ad8d/tests |
| 07:55 | <MikeSmith> | zcorpan: OK |
| 07:55 | <MikeSmith> | will try to get it set up today |
| 07:55 | <zcorpan> | thanks |
| 07:57 | <zcorpan> | yay sanity https://www.w3.org/Bugs/Public/show_bug.cgi?id=18521 |
| 07:58 | <annevk> | http://www.cs.brown.edu/research/plt/dl/domsemantics/domsemantics.pdf claims DOM4 is not complete with respect to Events |
| 07:58 | <annevk> | I wonder what they're referring to |
| 08:04 | <annevk> | zcorpan: mime type parsing is hard so we just gave up?! |
| 08:04 | <annevk> | IETF wins again |
| 08:06 | <zcorpan> | HOWTO makes APIs ignore your spec: make following your spec insanely complicated and have no gain for doing so |
| 08:08 | <AryehGregor> | smaug____, pong. |
| 08:11 | <AryehGregor> | hsivonen, currently I'm looking through the encoding spec's list of aliases and documenting which are supported by which browsers, preparatory to filing Gecko bugs for us to align with the spec where it makes sense (or spec bugs if it doesn't make sense). Does that sound useful? |
| 08:11 | <AryehGregor> | Or did someone already do something like thaT? |
| 08:11 | <AryehGregor> | that? |
| 08:12 | <AryehGregor> | (what bug component is this, and where do the aliases live in the code?) |
| 08:13 | <smaug____> | AryehGregor: dom tests. They live still in w3c, not whatwg? |
| 08:13 | <annevk> | AryehGregor: http://lxr.mozilla.org/mozilla-central/source/intl/locale/src/charsetalias.properties |
| 08:13 | <AryehGregor> | smaug____, whatwg has no place for tests right now, AFAICT. |
| 08:13 | <annevk> | AryehGregor: also http://lxr.mozilla.org/mozilla-central/source/intl/uconv/src/charsetData.properties |
| 08:13 | <AryehGregor> | The W3C doesn't seem to object to us keeping WHATWG-specific tests there, or at least no one's noticed. |
| 08:13 | <annevk> | AryehGregor: would be useful |
| 08:14 | <annevk> | AryehGregor: see also https://www.w3.org/Bugs/Public/show_bug.cgi?id=16773 |
| 08:17 | <hsivonen> | AryehGregor: seems useful |
| 08:17 | <AryehGregor> | Okay, good. |
| 08:18 | <hsivonen> | the charset menu in Firefox has way too much RDF for me to grok how it works |
| 08:38 | <annevk> | if I configured everything correctly anyone with access to github/whatwg can now maintain platform.html5.org |
| 08:38 | <annevk> | just push to https://github.com/whatwg/platform.html5.org |
| 08:57 | <annevk> | foolip++ |
| 08:57 | <annevk> | foolip: were you in the list of acknowledgments already? |
| 08:58 | <foolip> | annevk, I don't know, but it doesn't matter for a typo :) |
| 09:00 | <AryehGregor> | Does IE really not recognize "utf8" as an alias for "utf-8"? |
| 09:03 | <annevk> | foolip: I copy the style from Hixie_ where I don't try to value contributions and just acknowledge everyone that makes one |
| 09:04 | <foolip> | annevk, feel free to ack me if I'm not there, I will use it to brag about my awesome spec review powers :) |
| 09:04 | <annevk> | already done :) |
| 09:05 | <foolip> | annevk, btw, the Makefile doesn't work on a clean checkout, which is why I just changed both files |
| 09:05 | <foolip> | what's supposed to be in data? |
| 09:06 | <annevk> | foolip: added you as whatwg owner |
| 09:06 | <foolip> | annevk, cool, thanks :) |
| 09:07 | <annevk> | foolip: ah yeah, hg clone https://bitbucket.org/ms2ger/specification-data data |
| 09:07 | <annevk> | foolip: I guess we should move that over to whatwg/ too and add some details to the Makefile thingie |
| 09:08 | <annevk> | Ms2ger: ^^ |
| 09:09 | <foolip> | annevk, that'd be nice, yeah |
| 09:40 | <jgraham> | So, if I have a data URI that uses localStorage and then location.reload, why doesn't the storage area get recreated? |
| 09:43 | <AryehGregor> | annevk, so why in some cases did you require recognizing labels that only one major browser recognizes? E.g., all the iso88591/iso88592/etc. labels are recognized only by Gecko. |
| 09:43 | <AryehGregor> | (and are hard to read) |
| 09:44 | <AryehGregor> | Also koi8_r, and x-cp125*. |
| 09:44 | <AryehGregor> | And WebKit is the only one to recognize windows-949. |
| 09:44 | <zcorpan> | foolip: if you're so awesome at reviewing specs, why haven't you reviewed quirks mode? :-O |
| 09:46 | <jgraham> | In other news, why does the WHATWG even have "charter members"? If Hixie goes insane and the major stakeholders agreee change is needed then we would just fork the spec. No need for any shady-looky cabal at all. |
| 09:50 | <odinho> | zcorpan: The shock and horror! |
| 09:50 | <zcorpan> | inorite |
| 09:55 | <AryehGregor> | jgraham, has there been a single case where anyone has even used his role as a charter member? |
| 09:55 | <jgraham> | AryehGregor: I think they voted to allow annevk to also be a charter member |
| 09:55 | <AryehGregor> | Oh, well, that justifies the role's existence, then. |
| 10:00 | <odinho> | ^_^ |
| 10:01 | <annevk> | AryehGregor: also by Opera no? at least older versions of Opera implemented UTS22 |
| 10:01 | <annevk> | AryehGregor: WebKit has an open bug open for those too I think |
| 10:01 | <AryehGregor> | Well, the testing I was doing involved tedious manual labor, so I only tested IE/Gecko/WebKit. Sorry. :) |
| 10:03 | <AryehGregor> | # Currently .properties cannot handle : in key |
| 10:03 | <AryehGregor> | #iso_8859-4:1988=ISO-8859-4 |
| 10:03 | <annevk> | http://lists.w3.org/Archives/Public/www-archive/2012Apr/att-0058/spectable.html is my research fwiw |
| 10:03 | <AryehGregor> | Well, that explains that. |
| 10:03 | <annevk> | but just covers single-byte stuff |
| 10:03 | <AryehGregor> | Oh, nice. |
| 10:03 | <AryehGregor> | Ah, that's encoders/decoders? |
| 10:03 | <AryehGregor> | I'm testing .characterSet. |
| 10:04 | <annevk> | both labels and encoders/decoders |
| 10:04 | <annevk> | I would prolly pick a slightly different approach now, but back then I didn't know what I was looking at :-) |
| 10:10 | <AryehGregor> | Safari and Chrome are different? Is that because of different WebKit versions, or is this port-specific? |
| 10:10 | <AryehGregor> | (please tell me it's just a version difference) |
| 10:11 | <jgraham> | http://software.hixie.ch/utilities/js/live-dom-viewer/saved/1848 <- so Gecko and Opera seem to preserve localStorage in data urls across location.reload |
| 10:11 | <jgraham> | AFAICT WebKit doesn't allow localStorage in data urls at all |
| 10:12 | <AryehGregor> | <meta charset=utf-16> is ignored if it doesn't make sense, right? |
| 10:13 | <AryehGregor> | Actually, is there ever any point in saying <meta charset=utf-16>, or is the encoder sure to be able to figure it out? |
| 10:13 | <AryehGregor> | I.e., will it ever change the encoding used? Is it even legal to specify? |
| 10:15 | <zcorpan> | AryehGregor: iirc that means <meta charset=utf-8> |
| 10:15 | <AryehGregor> | Oh, really? |
| 10:15 | <AryehGregor> | But if the document actually seems to be encoded in UTF-16, is <meta charset> even effective, or is it just always ignored? |
| 10:16 | <zcorpan> | then it's ignored |
| 10:16 | <AryehGregor> | Figures. |
| 10:17 | <zcorpan> | "If charset is a UTF-16 encoding, change the value of charset to UTF-8." http://www.whatwg.org/specs/web-apps/current-work/multipage/parsing.html#prescan-a-byte-stream-to-determine-its-encoding |
| 10:18 | <zcorpan> | "If the new encoding is a UTF-16 encoding, change it to UTF-8." http://www.whatwg.org/specs/web-apps/current-work/multipage/parsing.html#change-the-encoding |
| 10:19 | <zcorpan> | (step 1 in #change-the-encoding aborts if it's already utf-16) |
| 10:20 | <zcorpan> | and the prescan looks at utf-16 BOM |
| 10:21 | <zcorpan> | i think the spec fails for BOMless utf-16 (without http-level decl) |
| 10:23 | <annevk> | AryehGregor: they are different |
| 10:23 | <annevk> | AryehGregor: Chrome has ICU with Google patches |
| 10:24 | <AryehGregor> | Blech. |
| 10:24 | <annevk> | AryehGregor: Safari uses Mac OS X ICU |
| 10:24 | <AryehGregor> | I usually just assume they're the same, but sometimes I get tripped up. |
| 10:25 | <annevk> | there's quite a bit that's different at the edges |
| 10:25 | <AryehGregor> | Yay. |
| 10:25 | <annevk> | e.g. URLs are different too |
| 10:25 | <annevk> | network too; most of the stuff I poke at is |
| 10:26 | <annevk> | jgraham: localStorage works for a unique origin? or is that a data URL in an <iframe>? |
| 10:28 | <jgraham> | In an iframe in this case. But I don't know why that would make a difference really |
| 10:28 | <jgraham> | You should have one storage area per origin |
| 10:32 | <annevk> | because then the origin is the same as the parent |
| 10:32 | <annevk> | except in WebKit which doesn't special case data URLs |
| 10:34 | <jgraham> | Where's that specified? |
| 10:34 | <jgraham> | I guess it makes sense, since it's also how they behave for scripts |
| 10:35 | <jgraham> | But it's not really obvious what the underlying mechanism is |
| 10:37 | <annevk> | AryehGregor: "Currently .properties cannot handle : in key" was fixed? |
| 10:38 | <annevk> | jgraham: in HTML under origin |
| 10:40 | <karlcow> | http://www.w3.org/mid/002e01cdab4c$198cbb50$4ca631f0$@tavis.ca |
| 10:40 | <karlcow> | that's good progress in the discussion |
| 10:40 | <jgraham> | annevk: I think the confusing is because "origin" is used in multiple senses |
| 10:41 | <jgraham> | annevk: Would be much clearer if the spec said "User agents must have a set of local storage areas, one for each document origin." |
| 10:42 | <jgraham> | But yes, I understand the intent now. In fact I think I did in the past but forgot again. Thanks |
| 10:42 | <annevk> | karlcow: kinda seems similar to the HTML parser arguments we have had countless of times... And the processor he suggests is just a validator and parser combined, not that special |
| 10:43 | <annevk> | karlcow: and as for the "can be done on top of RFC 3986" argument; 1. I've yet to see that 2. I don't think it's useful |
| 10:44 | <annevk> | jgraham: what trips me up is that origin is not exclusively tied to the URL |
| 10:44 | <annevk> | jgraham: every time |
| 10:45 | <karlcow> | annevk: not useful for your context, but seems useful for other users. worth trying to see if RFC3986 could be written in a way which has hooks to your documents and hoping that people propose actual text more than this should be done like this. That would be the second step. |
| 10:47 | <AryehGregor> | annevk, well, it compiles, and makes it pass my tests. I also glanced at the source code of the Python file that processes that, and didn't see how a : would make a difference. |
| 10:47 | <AryehGregor> | We'll see if it passes the try run. |
| 10:47 | <annevk> | karlcow: I'm not sure how it would be useful |
| 10:48 | <karlcow> | annevk: that's normal. It is not your context. |
| 10:48 | <annevk> | (again assuming we want a single processing library for URLs) |
| 10:48 | <annevk> | karlcow: saying it's normal doesn't help me understand |
| 10:49 | <AryehGregor> | annevk, maybe your spec could add the concept of parse errors and allow non-browser implementers to abort processing on the first parse error, like HTML does? |
| 10:49 | AryehGregor | hasn't been following at all, so maybe that doesn't make sense |
| 10:49 | <annevk> | AryehGregor: yeah, might do something like that |
| 10:49 | <hsivonen> | karlcow: to me, saying “do this on top of RFC xyz” is like saying “go fetch me a rock” |
| 10:50 | <hsivonen> | (in this case; not to suggest that to apply to all RFCs ever) |
| 10:50 | <karlcow> | hsivonen: I do not understand the expression "go fetch me a rock". |
| 10:50 | <AryehGregor> | annevk, it seems like "processing with all errors fatal" is necessarily a subset of "processing with detailed error recovery", and it would be a lot easier to include the former in a spec for the latter than vice versa. |
| 10:50 | <hsivonen> | karlcow: http://intertwingly.net/blog/2005/05/11/Fetch-Me-A-Rock |
| 10:51 | <annevk> | AryehGregor: yeah, just need to figure out what we want to make conforming |
| 10:51 | <karlcow> | hsivonen: then it is why I talked about contributions, more than discussions. |
| 10:51 | <AryehGregor> | hsivonen, how should the bugs for these be split up? All but windows-1252 and windows-1254 are already implemented by either IE or WebKit, so the compat risk should hopefully not be too large. https://bugzilla.mozilla.org/show_bug.cgi?id=802030#c2 |
| 10:52 | <annevk> | AryehGregor+=2 |
| 10:52 | <AryehGregor> | ? |
| 10:52 | <karlcow> | annevk: about normal/understand, that's true. The point of my comments was not about making someone understand. ☺ |
| 10:54 | <AryehGregor> | karlcow, why do you always have to come around and try to get us to play nice with the non-web people? We want to sneer at them for being impractical and ignore them, not be forced to acknowledge that there are implementations of Internet-related technologies that people care about other than web browsers. |
| 10:54 | <annevk> | AryehGregor: ++ seemed insufficient |
| 10:54 | <AryehGregor> | annevk, one ++ for each bug? |
| 10:54 | <jgraham> | ++++ |
| 10:54 | <annevk> | compiler error, nice one jgraham |
| 10:54 | <AryehGregor> | jgraham, is that valid C? |
| 10:54 | <jgraham> | I'm not sure |
| 10:54 | AryehGregor | doesn't see how it's ambiguous, strictly speaking |
| 10:54 | <karlcow> | AryehGregor: your *we* is not my *we* ☺ not interesting position. |
| 10:55 | <hsivonen> | AryehGregor: it might make sense to cover the cases that are already true in IE or WebKit in one bug and leave ISO-8859-1 as a special case bug |
| 10:55 | <jgraham> | Doesn't work in javascript |
| 10:55 | <annevk> | karlcow: look if all you are going to tell me that people have different view points, I'm not sure how that's helpful |
| 10:55 | <annevk> | karlcow: that seems pretty clear from the disconnect |
| 10:55 | <annevk> | karlcow: what I'd like is to get past the disconnect |
| 10:55 | <AryehGregor> | test.c:3:5: error: lvalue required as increment operand |
| 10:56 | <jgraham> | Yeah, same problem |
| 10:56 | <AryehGregor> | That's for "x++++". That makes sense. |
| 10:56 | <AryehGregor> | But then "++++x" should work? |
| 10:56 | <karlcow> | annevk: "get past the disconnect" ☺ yes. It goes both ways ;) |
| 10:56 | <AryehGregor> | No. Why not? |
| 10:56 | <AryehGregor> | Ah, I guess ++x and x++ both return rvalues. |
| 10:56 | <AryehGregor> | It's just that one is the value before increment and one after. |
| 10:58 | <annevk> | karlcow: well I'm glad you're having fun |
| 10:58 | <AryehGregor> | Oh, I misread my data. iso-8859-9 is the same as windows-1254 in WebKit too. |
| 10:58 | <AryehGregor> | hsivonen, and that bug should be extended to cover including us-ascii, too, since the spec says that should also be an alias for windows-1252. |
| 10:58 | <hsivonen> | AryehGregor: yeah |
| 11:00 | <hsivonen> | AryehGregor: so when you say gbk is the same as gb2312, do you mean the Encoding Standard is wrong when it makes gbk the preferred label? |
| 11:00 | karlcow | is not having fun nor being sad |
| 11:01 | <AryehGregor> | hsivonen, I mean that it treats gb2312 as an alias for gbk, while Gecko appears to treat them as distinct. |
| 11:01 | <AryehGregor> | (Vis-a-vis .characterSet, I dunno about encoders/decoders.) |
| 11:01 | <hsivonen> | AryehGregor: oh ok. |
| 11:02 | <hsivonen> | I expect Gecko to decode gb2312 the same way it decodes gbk, but I haven’t any proof at hand |
| 11:03 | <hsivonen> | one of these days I should implement the Encoding Standard in the Java version of the parser |
| 11:03 | <hsivonen> | (instead of using the alias facilities provided by the JDK) |
| 11:05 | <zcorpan> | TabAtkins: :past/:future are implemented in opera |
| 11:06 | <AryehGregor> | hsivonen, what's a bug I can use for a reference as to how to remove support for an encoding? |
| 11:10 | <zcorpan> | TabAtkins: i see 32 tests containing :past |
| 11:10 | <zcorpan> | (most seem to have "timestamp" in the file name) |
| 11:11 | <zcorpan> | TabAtkins: last i checked (at FOMS) nobody else implemented css for webvtt at all |
| 11:14 | <annevk> | AryehGregor: I think Gecko makes gb2312 a label for gb18030 but I'm not sure |
| 11:14 | <AryehGregor> | annevk, internally it's a label for GB2312, or so it seems. |
| 11:14 | <annevk> | AryehGregor: it has been suggested (see Encoding Standard bugs) to make all gb* a label for gb18030 |
| 11:15 | <annevk> | okay |
| 11:15 | <AryehGregor> | gb18030 is a separate internal encoding. |
| 11:15 | <AryehGregor> | I'm just looking at the labels, though, not encoders/decoders. |
| 11:16 | <Ms2ger> | AryehGregor, bug 623610 maybe? |
| 11:16 | <AryehGregor> | Ms2ger, what about it? |
| 11:16 | <AryehGregor> | Oh, I see. |
| 11:16 | <AryehGregor> | Thanks! |
| 11:16 | <Ms2ger> | Np |
| 11:31 | <annevk> | AryehGregor: the " Big5-HKSCS -> Big5" is prolly controversial for Gecko |
| 11:31 | <AryehGregor> | annevk, oh, really? Why? |
| 11:31 | <annevk> | Gecko has a completely custom HKSCS implementation |
| 11:31 | AryehGregor | suspects he's going to wind up with a giant patch that he has to rewrite from scratch if he tries to do them all at once |
| 11:31 | <annevk> | euh, completely custom big5 implementation |
| 11:32 | <annevk> | yeah, better to do them separate I suspect |
| 11:32 | <AryehGregor> | hsivonen, does it make sense to do all these merges at once, or should I try doing one first and see how it goes? |
| 11:32 | AryehGregor | hg revert -aC |
| 11:32 | <AryehGregor> | Which are the best ones to start with here? https://bugzilla.mozilla.org/show_bug.cgi?id=802082#c1 |
| 11:33 | <annevk> | Thai should be safe |
| 11:33 | <annevk> | ISO-8859-9 -> windows-1254 too |
| 11:33 | <annevk> | windows-949 -> euc-kr is just about adding a label right? |
| 11:33 | <AryehGregor> | Okay, so the three that convert to windows-*? |
| 11:33 | <AryehGregor> | No, they seem to be separate encodings. |
| 11:33 | <AryehGregor> | I have a separate bug for adding all the labels we want, no? |
| 11:33 | <annevk> | windows-874 is Thai |
| 11:33 | <annevk> | oh okay |
| 11:34 | <annevk> | yeah, the ones to windows-* are safe |
| 11:34 | <AryehGregor> | https://bugzilla.mozilla.org/show_bug.cgi?id=802059 |
| 11:34 | <AryehGregor> | Okay, I'll start with those. |
| 11:34 | <annevk> | ISO-8859-6 should be too, but email guys might have comments |
| 11:34 | <annevk> | and I guess the gbk one is safe, but see the Encoding Standard bugs, someone from Mozilla had some ideas |
| 11:37 | <foolip> | zcorpan, sorry, I only review specs with typos in them ;) |
| 11:38 | <zcorpan> | foolip: how do you know quirks doesn't have typos? |
| 11:39 | <jgraham> | If it had typos, he would have reviewed it |
| 11:39 | <jgraham> | ergo it doesn't |
| 11:40 | <zcorpan> | ah. ok. does that mean i rock? |
| 11:59 | <AryehGregor> | Merging two encodings is pretty tedious work. |
| 12:00 | <AryehGregor> | I'm glad I started with only three. |
| 12:00 | AryehGregor | wonders if it actually works |
| 12:00 | <AryehGregor> | This code is apparently shared for mail, so it might break mail stuff even if it's safe for the web. :( |
| 12:01 | <AryehGregor> | 41 files changed, 42 insertions(+), 1180 deletions(-) |
| 12:01 | AryehGregor | always likes patches like that |
| 12:06 | <jgraham> | Ones with 42 insertions? |
| 12:06 | <AryehGregor> | With 1138 net deletions. |
| 12:07 | <jgraham> | I know :p |
| 12:07 | <AryehGregor> | No, it was not a Hitchhiker's Guide reference. |
| 12:07 | <Ms2ger> | Not? |
| 12:09 | <annevk> | Are you sure? I write my specs just so that patches end up with references to The Guide |
| 12:11 | <annevk> | I wonder why Mail still allows anything but utf-8; seems so weird |
| 12:11 | <AryehGregor> | You mean for outgoing mail? |
| 12:12 | <annevk> | yes |
| 12:12 | <annevk> | btw, you might want to rename choices in the Encoding menu if you remove all variants |
| 12:13 | <hsivonen> | Is the “existing master branch” for the HTML WG documented somewhere? Who writes to it? |
| 12:13 | <annevk> | e.g. just have "Thai" |
| 12:13 | <AryehGregor> | Makes sense. |
| 12:13 | <AryehGregor> | Comment to the bug? |
| 12:14 | <annevk> | hsivonen: https://github.com/w3c/html/commits/master |
| 12:14 | <annevk> | AryehGregor: ok |
| 12:14 | <AryehGregor> | annevk, are you okay with this? https://bugzilla.mozilla.org/show_bug.cgi?id=801425#c12 |
| 12:16 | <annevk> | well no, but I don't really want to fight that with the SVG WG |
| 12:16 | <annevk> | so yes |
| 12:16 | <hsivonen> | so does a HTML 5.0 branch even exist in git yet? |
| 12:17 | <annevk> | hsivonen: btw, not a 100% sure that is it, but I think it is |
| 12:18 | <hsivonen> | what do “behind” and “ahead” mean here: https://github.com/w3c/html/branches ? |
| 12:18 | <jgraham> | hsivonen: AIUI master is the HTML 5 branch at the moment |
| 12:18 | <AryehGregor> | annevk, why isn't the use-case legitimate? Is there any better way to declaratively feature-detect in SVG? |
| 12:18 | <jgraham> | behind === missing commits in its upstream branch |
| 12:18 | <jgraham> | ahead === containing commits not in its upstream branch |
| 12:18 | <AryehGregor> | I don't know if the way they're doing it is ideal, but it looks like it meets a real use-case, unlike .hasFeature() for DOM. |
| 12:19 | <karlcow> | annevk: try |
| 12:19 | <karlcow> | defaults write com.apple.mail NSPreferredMailCharset "UTF-8" |
| 12:19 | <karlcow> | then restart mail.app |
| 12:19 | <jgraham> | Assuming github uses them in the same sense as git |
| 12:19 | <annevk> | AryehGregor: HTML doesn't have feature testing either; also it has all the same flaws as hasFeature() as dbaron mentions |
| 12:19 | <hsivonen> | jgraham: and upstream for feature/whatwg* is whatwg svn? |
| 12:20 | <annevk> | AryehGregor: and it's far too cumbersome to use |
| 12:20 | <SimonSapin> | HTML and XML attributes are not ordered, are they? |
| 12:20 | <annevk> | karlcow: not sure what you mean, I use Gmail at the moment and have it configured to always compose in utf-8 |
| 12:20 | <AryehGregor> | annevk, HTML has script. SVG is sometimes used in contexts where script isn't available, e.g., <img>. |
| 12:20 | <annevk> | SimonSapin: correct |
| 12:20 | <karlcow> | annevk: oooh my bad I thought you were talking about mail.app :) |
| 12:21 | <jgraham> | hsivonen: It looks like it means "compared to master", not "compared to upstream" |
| 12:21 | <SimonSapin> | I wonder what attr(*|foo) (in CSS) is supposed to mean with multiple attributes with that local name in different namespaces. |
| 12:21 | karlcow | has been fooled by the capitalization of Mail |
| 12:21 | <karlcow> | (pavlov) |
| 12:22 | <jgraham> | I don't really know why github doesn't have a view showing all the commits and branchpoints as a graph |
| 12:22 | <jgraham> | Or does it? |
| 12:22 | <hsivonen> | jgraham: do the changesets coalesce for the purposes of ahead/behind when the Editorial Team lands Hixie’s changesets out of order? |
| 12:22 | <Ms2ger> | SimonSapin, that's allowed? |
| 12:22 | <jgraham> | I don't know what happens when they land out of order, but I suspect that's one patch behind and one ahead |
| 12:23 | <jgraham> | Because if you cherry-pick a commit it gets a new SHA-1 |
| 12:23 | <hsivonen> | It annoys me that the chairs keep being coy about what they are up to instead of just saying what they are planning to do. |
| 12:23 | <annevk> | SimonSapin: if CSS allows that, sounds like a bug |
| 12:23 | <jgraham> | Well generally any rebasing gives you new SHA-1s |
| 12:23 | <SimonSapin> | Ms2ger: the grammar uses wqname, I guess we should change to qname |
| 12:23 | <Ms2ger> | Yay grammar |
| 12:24 | <hsivonen> | jgraham: right. what’s the best way to see the delta between the HTML WG and the WHATWG? |
| 12:24 | <jgraham> | hsivonen: I am not sure |
| 12:24 | <hsivonen> | what’s the purpose of all the different feature/whatwg* branches? |
| 12:24 | <jgraham> | Maybe compare commit messages? |
| 12:25 | <jgraham> | Or perhaps ask MikeSmith or darobin or nessy for help? |
| 12:26 | <MikeSmith> | as I understand it the purpose is to isolate changesets to make it possible to cherry-pick ones to merge into the W3C version |
| 12:26 | <MikeSmith> | but nessy could explain it much better |
| 12:27 | <MikeSmith> | or Robin |
| 12:27 | <MikeSmith> | I don't know what the best way is to get a diff |
| 12:27 | Philip` | notes randomly that Gerrit has a thing that automatically assigns a Change-Id to commits, which gets preserved when you cherry-pick or rebase etc, which makes it easier to track the relationships between different branches |
| 12:28 | <jgraham> | That sounds kind of cool |
| 12:28 | jgraham | wonders if Gerrit is any good |
| 12:29 | <Philip`> | Apparently it's less bad than many alternatives |
| 12:29 | <jgraham> | I mean it sounds like the closest thing that exists to the utterly-kickass custom code review tool we have |
| 12:29 | <jgraham> | I also heard it was impossible to set up |
| 12:30 | <Philip`> | The fun thing is trying to teach Perforce users how to use Gerrit and Git at the same time |
| 12:30 | <hsivonen> | Philip`! |
| 12:30 | <hsivonen> | Philip`: what’s your current outlook on fixing font subsetter bugs? |
| 12:30 | <jgraham> | Philip`: Wow, it sounds like your life ssucks ;) |
| 12:30 | <jgraham> | Also, it seems like it you come back here people start asking about bugfixes :) |
| 12:30 | <Philip`> | (Fortunately I'm not the one having to teach them :-) ) |
| 12:31 | <hsivonen> | Philip`: what should my expectations be wrt. https://bitbucket.org/philip/font-optimizer/issue/6/ligature-and-alternative-glyph ? |
| 12:32 | <Philip`> | hsivonen: I suppose realistically my current outlook is that I haven't cared about the font subsetter for many years and haven't got any immediate plans to start caring again |
| 12:32 | <Ms2ger> | Sounds like your canvas tests, then |
| 12:32 | <hsivonen> | Philip`: OK. thanks |
| 12:33 | <Philip`> | hsivonen: That sounds like it's mostly likely a bug in the code that rewrites the GSUB table |
| 12:33 | <hsivonen> | Philip`: yes. |
| 12:33 | <Philip`> | which is slightly scary code, if I remember correctly |
| 12:33 | <Philip`> | (since GSUB is slightly scary and has far too many features) |
| 12:33 | <hsivonen> | Philip`: I looked at that code and figured that in order to grok it, I’d have to read the spec for GSUB first |
| 12:34 | <Philip`> | Yeah, that's definitely required |
| 12:34 | <hsivonen> | Philip`: there was one comment about overriding a value to work around a bug in an underlying library |
| 12:34 | <hsivonen> | maybe I should just try removing that to see if the workaround is interacting badly with the library (which may or may not have been fixed by now) |
| 12:35 | <Philip`> | I think my usual approach for debugging the GSUB stuff was to use some tool that dumps the raw parsed GSUB data, and then look at it manually to see what looks dodgy |
| 12:36 | <Philip`> | Usually it's something boring like a set of N values being subsetted into a set of 0 values, and font libraries not liking that |
| 12:36 | <hsivonen> | At this point, I don’t know what I should try: fixing your code, figuring out if Google’s Java can be edited to do what I want, figuring out if Adobe’s Java can be edited to do what I want or trying to get fontforge Python API-based subsetting to work |
| 12:36 | <Philip`> | (and so you have to go up and delete whatever object is containing that set) |
| 12:36 | <hsivonen> | Philip`: do you have a recommendation for such a GSUB dump tool? |
| 12:37 | <Philip`> | I don't remember what I used :-( |
| 12:37 | <hsivonen> | OK. :-( |
| 12:37 | <Philip`> | It might have just been Font::TTF and Data::Dumper or something like that |
| 12:38 | <hsivonen> | I suppose eventually browsers will have to work around buggy fonts generated with your code |
| 12:38 | <hsivonen> | I already filed a bug against Safari... |
| 12:38 | <Philip`> | But I'm trying to start a habit of not being totally useless, so if you remind me some time this evening then I might be able to have a look at it myself (and try to find the old tools I was using) |
| 12:38 | <Philip`> | Do many people actually use my code? |
| 12:38 | <hsivonen> | I don’t know |
| 12:38 | Philip` | finds that frightening |
| 12:39 | <hsivonen> | It’s the most approachable option of all the alternatives I have found |
| 12:39 | <hsivonen> | Philip`: Google seems to use it |
| 12:39 | <Philip`> | I suppose I also ought to fix the many-years-old bug where it violates the Open Font License |
| 12:39 | <Philip`> | (Also, I suppose I should fix canvas tests, but that's too ambitious for now...) |
| 12:39 | <hsivonen> | Philip`: you mean including the Reserved Font Name in “subset of Foo” |
| 12:39 | <Philip`> | hsivonen: Yes |
| 12:40 | <hsivonen> | It’s a bit annoying that OFL and UFL require the opposite things on that point |
| 12:41 | Ms2ger | grumbles at bugzilla.opendarwin.org |
| 12:42 | <hsivonen> | Philip`: based on their logging of what they’ve done to particular fonts, it appears that Google Web Fonts admins use both subset.pl and their own subset.py that drives fontforge |
| 12:42 | <hsivonen> | but subset.py requires building fontforge from CVS, etc. |
| 12:44 | <Ms2ger> | Hm, the WHATWG bug widget still refers to "HTML5" |
| 12:45 | <Philip`> | Do you mean they use different tools for different fonts, or they use both on one font (getting the intersections of the subsets, I guess)? |
| 12:50 | <Philip`> | http://code.google.com/p/googlefontdirectory/source/browse/quando/METADATA?r=581d26fdfffeeff1f67f074ac12a7f65bdba8435 - not quite sure what that's trying to do |
| 12:50 | <Ms2ger> | annevk, oh, I guess you get to spec http://blogs.msdn.com/b/michkap/archive/2005/09/17/469941.aspx |
| 12:52 | <annevk> | Ms2ger: the spec covers that |
| 12:52 | <Ms2ger> | It does?! |
| 12:53 | <hsivonen> | Philip`: that particular case seems like using subset.pl for generating fonts for rendering the name of the font in a list of fonts and subset.py for generating a bunch of subsets for different reportoires for use on Web sites |
| 12:54 | <Ms2ger> | As in, rendering \ as � depending on charset? |
| 12:55 | <annevk> | Ms2ger: it doesn't cover fonts |
| 12:55 | <annevk> | Ms2ger: it does convert U+00A5 to U+005C during encoding for instance |
| 12:56 | <annevk> | Ms2ger: as he says, it's not dependent on the charset, it's something else |
| 12:56 | <Ms2ger> | Oh? |
| 12:57 | <Philip`> | hsivonen: Ah, right |
| 12:57 | <Philip`> | (Maybe the Fontforge thing is limited to Unicode blocks, you can't tell it precisely which characters to include?) |
| 12:57 | <Ms2ger> | Hm |
| 12:57 | <Ms2ger> | It's something weird |
| 12:57 | Ms2ger | ignores it for now |
| 12:57 | <hsivonen> | Philip`: I haven’t looked at the fontforge Python API |
| 12:58 | <hsivonen> | Philip`: it seems to me Google’s service design is limited to blocks |
| 12:59 | <Ms2ger> | http://dvcs.w3.org/hg/html/rev/27e1f424131b |
| 12:59 | <Ms2ger> | . . . |
| 12:59 | <AryehGregor> | . . . . . |
| 12:59 | <AryehGregor> | wat. |
| 13:11 | <AryehGregor> | annevk, okay, I'll update the hasFeature() spec now to return true except for stuff that looks like SVG. |
| 13:20 | <annevk> | Ms2ger: lolwut? |
| 13:49 | <AryehGregor> | Where does the data repo live these days? |
| 13:49 | <AryehGregor> | Build instructions for our specs would be handy. |
| 13:50 | <Ms2ger> | bitbucket still |
| 13:52 | <annevk> | AryehGregor: yeah, I'm planning on adding those |
| 13:53 | <AryehGregor> | Oh, right, there. |
| 13:53 | <annevk> | Ms2ger: you mind if I move it later? |
| 13:53 | <AryehGregor> | Ms2ger, hey, you're on github. How did that happen, hg-git? |
| 13:53 | <annevk> | Ms2ger: also, I was thinking if we just name it "data", we could change all the WHATWG Makefile's and make them assume it's one directory lower |
| 13:54 | <annevk> | Ms2ger: would work nicely |
| 13:54 | <annevk> | Ms2ger: or maybe "xref" or something |
| 13:54 | <AryehGregor> | Now where do I get anolis again? |
| 13:54 | <annevk> | AryehGregor: replace specification-data with anolis |
| 13:54 | <annevk> | i.e. same place |
| 13:55 | <annevk> | Ms2ger: then the Makefile simply says ../xref/ or some such |
| 14:13 | <Ms2ger> | annevk, mm, wfm, I guess |
| 14:13 | <Ms2ger> | AryehGregor, yes, hg-git |
| 14:16 | <Ms2ger> | AryehGregor, "Currently .properties cannot handle : in key" is a lie, btw |
| 14:17 | <AryehGregor> | Ms2ger, I noticed. |
| 14:17 | <AryehGregor> | Was it ever true? |
| 14:17 | <Ms2ger> | I believe so |
| 14:19 | <Ms2ger> | Henri Sivonen - Bug 563536, bug 204111 - Put data from uconv property files into code to avoid thread-unsafe property file loading. r=smontagu. |
| 14:19 | <Ms2ger> | Was what fixed it, I think |
| 14:35 | <GPHemsley> | jgraham: https://github.com/w3c/html/network |
| 14:36 | <GPHemsley> | (The Network tab on any repo shows the graph) |
| 14:46 | <jgraham> | GPHemsley: Possibly I don't understand how to use that, but it doesn't seem very useful |
| 14:46 | <jgraham> | Oh I see |
| 14:46 | <jgraham> | YOu have to drag it |
| 14:46 | <jgraham> | It still doesn't seem very useful |
| 15:04 | <smaug____> | div.dataset = {yo: 'yo', ma: 'ma', la: 'la'}; doesn't work, right? |
| 15:04 | <Ms2ger> | No |
| 15:04 | Ms2ger | looked at the same bug |
| 15:07 | <AryehGregor> | annevk_, so it looks like jQuery actually uses the DOM Level 3 Core .compareDocumentPosition behavior. In particular, that either DOCUMENT_POSITION_FOLLOWING or DOCUMENT_POSITION_PRECEDING is returned even for disconnected nodes, along with DOCUMENT_POSITION_IMPLEMENTATION_SPECIFIC. |
| 15:08 | <Ms2ger> | jQuery \o/ |
| 15:08 | <AryehGregor> | annevk_, to be honest, the API is horrible anyway and nobody in their right mind should use it for anything but sorting a list of nodes. |
| 15:08 | <AryehGregor> | For which, in fact, you need the crazy DOM Level 3 Core behavior. |
| 15:08 | <AryehGregor> | And which is what jQuery uses it for. |
| 15:08 | <AryehGregor> | It's not a terrible use-case, the API is just amazingly bad. |
| 15:09 | <AryehGregor> | IE, Gecko, and Opera have interop on returning FOLLOWING/PRECEDING even for disconnected nodes. |
| 15:09 | <AryehGregor> | So I think the right thing here is probably to spec it. |
| 15:10 | <hsivonen> | was http://www.w3.org/TR/widgets-uri/ abandoned because of general disinterest in Widgets, because of getting fed up with battling the TAG or because of disinterest from implmentors who were nonetheless implementing some Widget specs? |
| 15:10 | hsivonen | just proposed using widget: for addressing into EPUB books in B2G... |
| 16:04 | <AryehGregor> | Ms2ger, do you mind if I steal your hasFeature tests, since they no longer match the spec and I'm about to write new ones? |
| 16:04 | <AryehGregor> | I figured I'd base them on yours. |
| 16:04 | <AryehGregor> | Just to have a pool of likely strings to test. |
| 16:05 | <AryehGregor> | (although, they're all supposed to return true anyway) |
| 16:40 | <magcius> | Is http://www.w3.org/TR/css3-syntax/ still relevant? |
| 16:41 | <magcius> | Or is http://www.w3.org/TR/CSS2/syndata.html more up to date? |
| 16:41 | <magcius> | According to the CSS3 grammar, it's impossible to have signed numbers. |
| 16:44 | <magcius> | But the CSS 2.1 grammar wouldn't allow '-vendor-prefix' things, because otherwise it would be parsed as an ident. |
| 16:44 | <magcius> | Does the tokenization step now depend on context? |
| 17:14 | <Ms2ger> | AryehGregor, of course, go ahead |
| 17:44 | <hsivonen> | Philip`: did your memory about ttf debugging tools refresh? |
| 18:01 | <hober> | hsivonen: you could use EPUB CFI or whatever it's called |
| 18:09 | <TabAtkins> | magcius: The TR version of Syntax is irrelevant and old and busted (check the date on it - 2003!). I'm actively working on the ED version of it at http://dev.w3.org/csswg/css3-syntax/ |
| 18:10 | <magcius> | TabAtkins, aha, thanks! |
| 18:10 | <TabAtkins> | The 2.1 grammar *has* been changed to allow +/- directly in the number token, but it's in errata, not in the main document. |
| 18:10 | <TabAtkins> | As written in 2.1 currently, +/- have to be explicitly requested as DELIM token whenever you use a NUMBER token (or things that build on NUMBER) in a grammar. |
| 18:11 | <TabAtkins> | The new Syntax spec implements the resolution to have +/- be part of the NUMBER token directly. |
| 18:12 | <magcius> | TabAtkins, that's amazing. Thank you so much! |
| 18:13 | <magcius> | TabAtkins, I'm loving this new spec. |
| 18:13 | <magcius> | Shiny with the railroad diagrams |
| 18:13 | <TabAtkins> | Yus! |
| 18:19 | <hober> | I wonder who will chair the Web Notification WG now. |
| 18:40 | <Ms2ger> | hober, you |
| 18:42 | <hober> | Ms2ger: no thanks |
| 18:42 | <TabAtkins> | No backsies! |
| 18:43 | <hober> | :) |
| 18:43 | <gsnedders> | AryehGregor: Try googling Anolis :) |
| 19:08 | <hsivonen> | hober: what's EPUB CFI and use it for what? |
| 19:10 | <hsivonen> | hober: no, it doesn't fit |
| 19:11 | <hsivonen> | it's not a hierachical URL scheme for the contents of the EPUB archive |
| 19:19 | <hober> | hsivonen: ahh, ok |
| 19:20 | <annevk_> | hober: mutation events are dead |
| 19:20 | <annevk_> | hober: mutation observers are not |
| 19:20 | <annevk_> | AryehGregor: man, compareDocumentPosition() blows |
| 19:21 | <annevk_> | http://lists.w3.org/Archives/Public/www-archive/2012Oct/0064.html filed for future claim chowder? well not really, but I wish we kept a list of those somewhere |
| 19:40 | <hober> | annevk_: I knew I'd mention whichever one was the old one :) |
| 19:43 | <zewt> | would have been better called "change events" or something heh |
| 19:43 | <TabAtkins> | I think we explicitly got away from the "events" name because you *don't* get events for individual changes. |
| 19:44 | <zewt> | the problem is "mutation" |
| 19:44 | <TabAtkins> | You get an event "as soon as possible" with a list of everything that's changed since your last event. |
| 19:44 | <zewt> | since everyone remembers "mutation something" and then forgets which is which, heh |
| 19:44 | <TabAtkins> | True that. |
| 20:07 | <gsnedders> | annevk_: Can we add a new compareDocumentPosition function? |
| 20:08 | <gsnedders> | annevk_: where < 0 when preceding, > 0 following, 0 when equal. Maybe make it -1 if it's an ancestor, -2 if not, or something? |
| 20:08 | <gsnedders> | Same with children. |
| 20:10 | <TabAtkins> | Like a proper comparator, taking two Nodes as arguments? |
| 20:10 | <gsnedders> | Indeed. |
| 20:10 | <TabAtkins> | I support. |
| 20:11 | <gsnedders> | Or even foo.compare(bar) |
| 20:11 | <gsnedders> | Doesn't matter which model is used, really. |
| 20:11 | <TabAtkins> | Nah, needs to be a real comparator so you can pass it to []#sort. |
| 20:11 | <TabAtkins> | Or both, whatever. |
| 20:11 | <gsnedders> | Yeah, guess that's better. |
| 20:12 | <gsnedders> | Instead of requiring (x, y) => x.compare(y) or whatever |
| 20:12 | <TabAtkins> | If we're embedding information in the side-channel of magnitude, I find it useful to know if something is an ancestor, a previous sibling, a previous sibling of an ancestor, or unrelated-but-preceding. |
| 20:12 | <TabAtkins> | Yeah. |
| 20:13 | <gsnedders> | Yeah, I'd rather have magnitude differ my something like that. |
| 20:13 | <TabAtkins> | Though I suppose just "ancestor" is the cheapest one to answer, since you obtain the information automatically just by performing the check. |
| 20:14 | <gsnedders> | Is sibling of an ancestor that important? |
| 20:14 | <TabAtkins> | It's useful for the CSS counter algorithm, so I suspect it might be useful for other things with similar structure. |
| 20:14 | <TabAtkins> | (which reminds me - I need to rewrite the counter algorithm today again) |
| 20:15 | <gsnedders> | Really? I thought the algorithm only cares about document order. |
| 20:15 | <TabAtkins> | Nope, you inherit scopes from your previous sibling, or your ancestor. You then inherit *values* from the immediately preceding element in document order. |
| 20:16 | <TabAtkins> | Depending on how you write the algorithm, it's not strictly necessary to know about siblings of ancestors. |
| 20:16 | <gsnedders> | Oh, isn't the scope just taken by the parent element of where it's defined? That's what I always think of it as? |
| 20:16 | <TabAtkins> | Nope, you can get scopes from your siblings. |
| 20:16 | <TabAtkins> | If they *establish* one. |
| 20:17 | <gsnedders> | Right, I just think of it going to the parent of where it is established, but not applying to that element directly. |
| 20:17 | <gsnedders> | Which is effectively the same thing. |
| 20:18 | <TabAtkins> | No, it doesn't go to the parent. The parent can establish the same counter, and that gives you two nested counters, the innermost coming from your sibling. |
| 20:18 | <TabAtkins> | Plus it doesn't apply to preceding siblings. |
| 20:19 | <TabAtkins> | I guess figuring out siblings is easy from the comparison algorithm, too - it's "yes" if the comparison ended after a single walk up from both. |
| 20:19 | <gsnedders> | Looking over this, the spec does match my model, so I guess I'm just explaining it badly. |
| 20:19 | <TabAtkins> | I think the spec as written is wrong, since I wrote it recently and am pretty sure I got it wrong. |
| 20:21 | <gsnedders> | What I think matches what you're saying here, basically, too. |
| 20:22 | <TabAtkins> | Um, okay. ^_^ |
| 20:22 | <TabAtkins> | The clearest statement of what I mean is the "inherit scopes from previous sibiling, or else parent; inherit values from document-preceding element". |
| 20:22 | <gsnedders> | Basically I have a different model of it in my head that's semantically equiv. |
| 20:22 | <TabAtkins> | That's entirely correct. |
| 20:23 | <gsnedders> | More complicated edge-cases, but makes me reason about the normal case more easily. |
| 21:04 | <hober> | annevk_: https://www.w3.org/Bugs/Public/show_bug.cgi?id=17057#c3 |
| 21:07 | Philip` | keeps typoing "hg" as "git" :-( |
| 21:07 | <Philip`> | hsivonen: Looks like the problem with that font is the "# XXX: This is a really horrid hack." part |
| 21:08 | <Philip`> | If it's commented, does that mean it's a feature not a bug? |
| 21:10 | <Philip`> | The easy solution is to add a "return 0" in that bit, which makes it drop the whole lookup group thing, and then all the other ligatures become visible |
| 21:10 | <Philip`> | but I don't know whether the data it's dropping is important for anything |
| 22:15 | <SimonSapin> | TabAtkins: (re www-style) Ok, great :) |
| 22:15 | <TabAtkins> | SimonSapin: ^_^ |
| 22:17 | <SimonSapin> | TabAtkins: how much can we change stuff that is already in CR? Were these ok because they are minor? |
| 22:18 | <TabAtkins> | Depends. We *can* change anything, but some types of changes require a quick circle back to LC for Process reasons. |
| 22:18 | <TabAtkins> | I think we're allowed to make changes in the transition to PR? I dunno. |
| 22:18 | <TabAtkins> | fantasai would know better. |
| 22:18 | <SimonSapin> | How do we know which is which? |
| 22:19 | <TabAtkins> | Things which make implementations of the old spec invalid require an LC pass. |
| 22:19 | <TabAtkins> | Doing the opposite is fine. |
| 22:19 | <TabAtkins> | And editorial changes are fine. |
| 22:20 | <SimonSapin> | got it, thanks |
| 22:21 | <SimonSapin> | (isn’t wqname -> qname in the first case?) |
| 22:21 | <TabAtkins> | Yes. |
| 22:21 | <TabAtkins> | (Though going the from qname -> wqname would be the second case - loosening a restriction.) |
| 22:22 | <TabAtkins> | Conceptually, it's just whether a complete implementation of the old spec would now be incomplete (okay) or incorrect (not okay). |
| 22:23 | <SimonSapin> | I’ve only tested three implementations, but none of them parsed attr(*|foo) |
| 22:24 | <TabAtkins> | Right, that change happened during LC, so it's likely that impls arent' updated to it yet. |
| 22:24 | <TabAtkins> | But the rule is based on a *theoretical* complete implementation. |
| 22:26 | <SimonSapin> | so do we need one more cycle for css3-values for today’s changes? |
| 22:27 | <TabAtkins> | We will, yeah. Probably do it after the testsuite is done, when we're about ready to do PR. |
| 22:28 | <TabAtkins> | So as to avoid cycling through multiple times. |
| 22:48 | <Philip`> | hsivonen: Ah, looks like the offending GSUB subtable is used for e.g. "ch" when you enable dlig |
| 22:48 | <Philip`> | so it seems like it'd be nicest to handle that case correctly |
| 22:50 | Philip` | tries to remember enough of how all the code works |
| 22:58 | Philip` | decides it's a problem for tomorrow, not today |
| 23:17 | <jamund> | I want to update http://wiki.whatwg.org/wiki/Validator.nu_Web_Service_Interface and add a CORS example |
| 23:17 | <jamund> | https://gist.github.com/3902535 |
| 23:18 | <Hixie_> | do you need an account? |
| 23:18 | <jamund> | Sure. I don't know how to get access. |
| 23:18 | <Hixie_> | tell me your username, name, e-mail address |
| 23:18 | <jamund> | xjamundx, Jamund Ferguson, jamund⊙gc |
| 23:19 | <Hixie_> | it is done, check e-mail |
| 23:20 | <jamund> | thanks. |