| 00:55 | <jarib> | z~ |
| 01:25 | <MikeSmith> | are there any command-line utilities that, given a file, will report the character encoding of the file? |
| 01:27 | <jcranmer> | file, maybe? |
| 01:35 | <MikeSmith> | jcranmer: indeed |
| 01:35 | <MikeSmith> | thanks |
| 01:36 | MikeSmith | did not know about the --mime-encoding switch in file |
| 01:55 | <Hixie> | Philip`: http://canvex.lazyilluminati.com/survey/2007-07-17/analyse.cgi/pages/tagattr/img/longdesc 500 |
| 02:54 | <MikeSmith> | file command doesn't seem to know the difference between a us-ascii file and a utf-8 one |
| 03:05 | <wirepair> | file -bi doesn't work? |
| 03:07 | MikeSmith | tries file -bi |
| 03:08 | <MikeSmith> | wirepair: nope |
| 03:08 | <wirepair> | and isn't us-ascii part of utf-8? heh |
| 03:08 | <jcranmer> | presumably it can't tell if there are no high-bit characters |
| 03:09 | <wirepair> | yeah or there's no BOM |
| 03:10 | <MikeSmith> | validator.nu seems to be smarter - http://validator.nu/?doc=http%3A%2F%2Fwww.w3.org%2FTR%2Fhtml-markup%2F |
| 03:10 | <MikeSmith> | too smart for its own damn good |
| 03:13 | <wirepair> | no it's not |
| 03:13 | <wirepair> | it's reading the servers charset tag |
| 03:13 | <wirepair> | ;> |
| 03:14 | <wirepair> | Content-Type: text/html; charset=utf-8 |
| 03:14 | <wirepair> | that's why it disagree's i imagine |
| 03:14 | <wirepair> | er disagrees |
| 03:14 | <wirepair> | i see no bom nor non-ascii codes in that html file |
| 03:14 | <wirepair> | only reference to utf-8 is the server's content-type header |
| 03:26 | <MikeSmith> | wirepair: ah |
| 03:36 | <wirepair> | just my guess though ;> |
| 03:37 | <MikeSmith> | wirepair: I'm not familiar with that part of the validator.nu |
| 03:38 | <wirepair> | oh duh it's right there |
| 03:38 | <wirepair> | under parser |
| 03:38 | <wirepair> | says parser: 'automatically from content-type' |
| 03:38 | <MikeSmith> | but the fact that hsivonen has is reporting, "disagrees with the actual encoding of the document", made me think that he must have it doing something to actually check the encoding, not just relying on the headers |
| 03:39 | <MikeSmith> | wirepair: I think that only is used for deciding whether to parse is as text/html or as XML |
| 03:39 | <wirepair> | gotcha |
| 03:39 | MikeSmith | makes note to ask hsivonen about it when he gets back |
| 03:40 | <wirepair> | that site isn't very helpful |
| 03:40 | <MikeSmith> | which site? |
| 03:40 | <wirepair> | validator |
| 03:40 | <MikeSmith> | what's not useful about it? |
| 03:40 | <wirepair> | the fact that it doesn't tell you where it decided it matches utf-8 |
| 03:40 | <wirepair> | and not ascii |
| 03:41 | <wirepair> | because i'm looking at the raw bytes, and i see no multibytes just straight up ascii heh |
| 03:41 | <MikeSmith> | yeah, it probably could provide more details |
| 03:41 | <MikeSmith> | I guess I should just go ahead and look at the htmlparser code and see for myself |
| 03:43 | <MikeSmith> | hmm, there's not "ack" macports port, it seems |
| 03:44 | MikeSmith | falls back to find . | xargs grep |
| 03:49 | <MikeSmith> | wirepair: http://hg.mozilla.org/projects/htmlparser/file/546412142175/src/nu/validator/htmlparser/io/Driver.java |
| 03:49 | <MikeSmith> | in case you're curious |
| 03:50 | <MikeSmith> | it seems that at the point where it emits that error, it hasn't recorded how it actually determined the encoding |
| 03:50 | MikeSmith | looks at getActualHtmlEncoding() |
| 03:56 | <wirepair> | found something? |
| 03:59 | <MikeSmith> | it's using InputSource.getEncoding() |
| 03:59 | <MikeSmith> | from org.xml.sax |
| 04:00 | <MikeSmith> | http://java.sun.com/j2se/1.5.0/docs/api/org/xml/sax/InputSource.html#getEncoding() |
| 04:00 | <MikeSmith> | I have no clue what means getEncoding actually uses to determine the encoding |
| 04:01 | <wirepair> | yeh,, no idea heh |
| 04:07 | <MikeSmith> | actually, it's just a simple getter |
| 04:07 | <wirepair> | wtf |
| 04:07 | <wirepair> | heh |
| 04:08 | <wirepair> | so something is supplyin gthe stream with an encoding ? |
| 04:08 | <wirepair> | kinda makes the whole validation thing moot doesn't it? |
| 04:08 | <MikeSmith> | the magic is here: http://hg.mozilla.org/projects/htmlparser/file/546412142175/src/nu/validator/htmlparser/io/HtmlInputStreamReader.java |
| 04:09 | <wirepair> | i didn't see a BOM so... |
| 04:09 | <wirepair> | again, would be better if the validator says where it no longer matches the expected type |
| 04:10 | <MikeSmith> | It seems to be using jchardet and/or ICU4J |
| 04:10 | <wirepair> | so people wouldn't have to look through the validator code to figure out how it's determining hehe |
| 04:10 | <MikeSmith> | true |
| 04:10 | <wirepair> | i say just wait ;> |
| 04:12 | <MikeSmith> | well, at this point, I'm reasonably sure it is actually doing genuine character-encoding detection, not just relying on the Content-Type header |
| 04:12 | <MikeSmith> | actually, I already knew that, because there are times when it will report that the character-encoding in the Content-Type header doesn't match the actual encoding of the document |
| 04:13 | <MikeSmith> | anyway, I need to get some grub |
| 04:13 | <MikeSmith> | wirepair: btw, you have a lot of sento in Oota-ku |
| 04:14 | <MikeSmith> | a number of which are real onsen |
| 04:14 | <MikeSmith> | well, real onsen water |
| 04:14 | <wirepair> | ah not much of an onsen person |
| 04:14 | <MikeSmith> | http://www21.ocn.ne.jp/~spa-mich/todofuken/tokyo/tokyo.htm |
| 04:14 | <MikeSmith> | I am nuts for onsen |
| 04:15 | <wirepair> | ha |
| 04:15 | <wirepair> | i'll show the wife |
| 04:15 | <wirepair> | thanks ;> |
| 04:16 | <MikeSmith> | yeah, for some reason you all have more onsen-sento there than any other part of the city, by a wide margin |
| 04:17 | <wirepair> | weird.. |
| 04:17 | <wirepair> | wonder why that is |
| 04:31 | <MikeSmith> | wirepair: I think one reason might be that Oota-ku is just really big |
| 04:31 | <MikeSmith> | geographically |
| 04:48 | <wirepair> | how long you been in japan mike? |
| 04:51 | <MikeSmith> | wirepair: 8 years |
| 04:51 | <wirepair> | wow, quite a bit of time ;> |
| 04:51 | <wirepair> | 2x me hehe |
| 04:51 | <wirepair> | i imagine your japanese must be pretty damn good by now hehe |
| 04:54 | <MikeSmith> | wirepair: my Japanese is actually pretty bad |
| 04:54 | <wirepair> | didn't study or gave up studying? |
| 04:54 | <wirepair> | hehe |
| 04:55 | <MikeSmith> | I studied some, but mostly learned on the job and from my daughter, who's 12 years old now |
| 04:55 | <MikeSmith> | she and I only speak Japanese to each other |
| 04:56 | <wirepair> | interesting |
| 04:56 | <MikeSmith> | she understands a lot of English, but doesn't speak English much |
| 04:56 | <wirepair> | yeah i've seen that, where the child understand what you say, but will just answer in japanese |
| 04:56 | <MikeSmith> | yeah |
| 04:57 | <MikeSmith> | she is starting to study English more, and will be studying it at school when she starts junior-high in April |
| 04:57 | <wirepair> | crazy |
| 04:57 | <MikeSmith> | she teaches me a lot of words, actually |
| 04:57 | <wirepair> | i imagine her pronounciation will be better than most |
| 04:57 | <wirepair> | hehe |
| 04:58 | <wirepair> | funny i was teaching a family in laws kid a little english (he's like 3) and his father was there, i said truck in english pronounciation, and his father said it in japanese |
| 04:58 | <wirepair> | the kid was like 'which is it?!' |
| 04:58 | <MikeSmith> | heh |
| 04:58 | <wirepair> | he thought they were completely different words |
| 04:59 | <MikeSmith> | yeah, it's like orange, or cake, or many others |
| 04:59 | <MikeSmith> | truck is a tough one, though |
| 04:59 | <wirepair> | yup |
| 05:00 | <MikeSmith> | "tr" is just not amenable to katakana-ized pronunciation |
| 05:00 | <wirepair> | chi-a-ru |
| 05:00 | <wirepair> | hehe |
| 05:01 | <MikeSmith> | but I think it's good for kids to learn that people pronounce the same words in different ways |
| 05:01 | <wirepair> | yeah definitely |
| 05:01 | <MikeSmith> | it's definitely confusing when they are young, though |
| 05:03 | <MikeSmith> | one example is the way some Japanese teachers of English in secondary schools pronounce the word "the" |
| 05:03 | <wirepair> | za-? |
| 05:03 | <MikeSmith> | which is like "za" |
| 05:03 | <wirepair> | yeah |
| 05:03 | <MikeSmith> | yeah |
| 05:04 | <MikeSmith> | some people think it's just wrong to teach kids to pronounce it that way |
| 05:04 | <wirepair> | i don't think too highly of the japanese educational system to be honest though |
| 05:04 | <wirepair> | not from the things i've heard |
| 05:04 | <MikeSmith> | yeah, the system has some problems, for sure |
| 05:04 | <MikeSmith> | but about particular case of "za", I don't think it's a big deal at all |
| 05:05 | <MikeSmith> | because for one thing, it's also the way some European non-native-English speakers say it soo |
| 05:05 | <wirepair> | yeah, i'm more concerned with my child sitting for 8hours and then being expected to regurgitate the information back |
| 05:05 | <wirepair> | ;> |
| 05:05 | <wirepair> | not that i have any at the moment, but when i do heh |
| 05:06 | <MikeSmith> | well, my own secondary education wasn't exactly a thrill ride |
| 05:07 | <wirepair> | at the time i thought mine was terrible, looking back and hearing other peoples experiences i guess it was pretty damn good |
| 05:07 | <wirepair> | heh |
| 05:07 | <MikeSmith> | I had teachers who were supposedly better trained and more enlightened, but still most of them sucked at actually teaching |
| 05:07 | <wirepair> | i only had a few like that |
| 05:07 | <MikeSmith> | teaching well is a rare talent |
| 05:07 | <wirepair> | indeed. |
| 05:08 | <MikeSmith> | I taught high school on contract for 3 years or so |
| 05:08 | <wirepair> | ah so you got to see it firsthand ;> |
| 05:09 | <MikeSmith> | well, it was enough for me to realize I couldn't do it well, and needing to find some other way to earn money |
| 05:09 | <wirepair> | ahha |
| 05:09 | <wirepair> | gotcha |
| 05:09 | <MikeSmith> | anyway, in 6 years of primary school, my daughter has had some pretty good teachers, and some pretty bad ones |
| 05:09 | <MikeSmith> | the teachers do make a big difference |
| 05:10 | <wirepair> | yeah |
| 05:10 | <MikeSmith> | last year, her teacher was a young-guy, first-year teacher, and he didn't really know what he was doing, and her grades went down |
| 05:10 | <wirepair> | ahh that's a shame |
| 05:11 | <MikeSmith> | this year, she had a much older, more experienced, no-nonsense teacher and she learned a lot more and turned all her grades around |
| 05:12 | <MikeSmith> | I know there are some good teachers in secondary schools here, too |
| 05:12 | <MikeSmith> | along with mediocre and bad ones |
| 05:13 | <MikeSmith> | it's partly just the luck of the draw, I guess |
| 05:13 | <wirepair> | yeah i'd like to know what the curriculum is like, my understanding is there's little to know real teaching of the scientific process until college |
| 05:13 | <wirepair> | s/know/no |
| 05:14 | <wirepair> | but hey what do i know, i never even went to college ;> |
| 05:14 | <MikeSmith> | I think the system expects -- or even requires -- a lot more personal responsibility on the part of students and their parents |
| 05:15 | <wirepair> | yeah, unfortunately what i've seen here is an unbelievable amount of pressure on teachers to take care of kids |
| 05:15 | <MikeSmith> | the thing I value most about my college experience was the many opportunities it provided for relaxed dope smoking |
| 05:15 | <wirepair> | hahah |
| 05:16 | <wirepair> | see i did that in private school for 2 years, i got it out of my system |
| 05:16 | <wirepair> | then i got a job during the bubble, it bust, and all my friends got out of college and couldn't find jobs when i had 4 years experience ;> |
| 05:17 | <wirepair> | my timing was impeccable |
| 05:17 | <MikeSmith> | you planned that well |
| 05:18 | <wirepair> | just very lucky |
| 07:12 | <othermaciej> | MikeSmith: check out the status on this: http://www.w3.org/TR/2010/WD-html-bidi-20100304/#bidi-isolation |
| 07:12 | <othermaciej> | MikeSmith: "This document is an editors' copy that has no official standing." |
| 07:13 | <othermaciej> | that's probably not right |
| 07:13 | <MikeSmith> | oops |
| 07:13 | <othermaciej> | MikeSmith: also I surprisingly hadn't heard about that document |
| 07:13 | <MikeSmith> | I will fix it in-place now |
| 07:13 | othermaciej | wonders why the proposals in that document weren't proposed to the HTML WG and hopes they will be at some point |
| 07:14 | <MikeSmith> | othermaciej: I probably should not admit this, but the first I saw of that doc was this announcement |
| 07:15 | <MikeSmith> | I think Richard mentioned something about it to me, but it didn't register as being something closely related to HTML |
| 07:15 | <othermaciej> | glad it's not just me |
| 07:16 | <othermaciej> | Looking at the TransReq Philippe did say it should be be brought to the attention of the HTML WG |
| 07:17 | <MikeSmith> | well |
| 07:18 | <MikeSmith> | it's odd that they chose to publish it as WD first |
| 07:18 | <MikeSmith> | the "bring to attention of the HTML WG" step would seem like it should have better occurred first |
| 07:18 | <MikeSmith> | wait |
| 07:19 | <MikeSmith> | it's about bidi |
| 07:19 | <MikeSmith> | so they are doing everything backwards |
| 07:19 | <MikeSmith> | sorry, not backwards, just different-wards |
| 07:19 | <MikeSmith> | it's all relative |
| 07:23 | <MikeSmith> | hsivonen: btw, I wanted to ask you about the character-detection mechanism that validator.nu uses |
| 07:24 | <MikeSmith> | hsivonen: http://validator.nu/?doc=http%3A%2F%2Fwww.w3.org%2FTR%2Fhtml-markup%2F |
| 07:24 | <MikeSmith> | "Internal encoding declaration us-ascii disagrees with the actual encoding of the document (utf-8)" |
| 07:25 | <MikeSmith> | but that document actually has no non-ascii characters in it, and no BOM |
| 07:26 | <MikeSmith> | so I assume the character-detection code is actually also using the Content-type header |
| 07:27 | <MikeSmith> | not just checking the "actual encoding" of the document |
| 07:28 | <MikeSmith> | if so, I'm wondering if that error message might be refined to say, "disagrees with the actual encoding of the document and/or the Content-Type header the document is being served with (utf-8)" |
| 07:28 | <MikeSmith> | or something similar |
| 07:30 | <hsivonen> | I guess the actual encoding is the one named on the http level in this case |
| 07:30 | <zcorpan> | MikeSmith: pong |
| 07:31 | <MikeSmith> | hsivonen: OK |
| 07:31 | <hsivonen> | maybe 'actual' is a philosophical/onntological question and the message should say something else |
| 07:31 | <othermaciej> | heh, bidi |
| 07:32 | <MikeSmith> | hsivonen: yeah, the word "actual" is what threw me |
| 07:32 | <hsivonen> | in this case, actual means the encoding that's actually used for decoding |
| 07:32 | <MikeSmith> | ah |
| 07:32 | <MikeSmith> | that's more like maybe "effective" |
| 07:33 | <MikeSmith> | I can't think of any great suggestions other than that one |
| 07:33 | <MikeSmith> | which is not great |
| 07:33 | <MikeSmith> | anyway, I'm not sure it's worth changing |
| 07:33 | <MikeSmith> | I had just been wondering |
| 07:34 | <MikeSmith> | zcorpan: I wanted to tell you I made some more changes to the markup doc based on the errors you reported a while back |
| 07:35 | <MikeSmith> | zcorpan: http://www.w3.org/TR/html-markup/syntax.html#doctype-syntax |
| 07:36 | <MikeSmith> | my rationale for organizing the description of doctypes that way (instead of the way Hixie has it) is that I wanted to try to have the description of what a "normal" HTML doctype is to be as simple as possible |
| 07:37 | <MikeSmith> | I don't know if "normal doctype" is the best choice as far a terminology |
| 07:38 | <MikeSmith> | but anyway, I did correct the description of "deprecated doctype" to remove any reference to the XML definition of what a docytpe is |
| 07:38 | <MikeSmith> | http://www.w3.org/TR/html-markup/syntax.html#deprecated-doctype |
| 07:39 | <annevk> | the piece on W3C news about H:TML says more about what it is not than what it is :/ |
| 07:39 | <MikeSmith> | heh |
| 07:39 | <zcorpan> | MikeSmith: http://www.w3.org/Bugs/Public/show_bug.cgi?id=9145 |
| 07:40 | <MikeSmith> | annevk: my fault, since it's word-for-word from the abstract |
| 07:40 | <MikeSmith> | zcorpan: looking now |
| 07:40 | <MikeSmith> | OK |
| 07:41 | MikeSmith | makes that same change now |
| 07:42 | <MikeSmith> | hsivonen: "computed encoding"? |
| 07:42 | <zcorpan> | MikeSmith: "Conditionally, depending on whether it is part of a permitted-public-ID-system-ID-combination, the following parts, in exactly the following order:" doesn't say whether the condition is true or false |
| 07:42 | <zcorpan> | MikeSmith: maybe "If it (what?) is part of a permitted-..." |
| 07:42 | <MikeSmith> | yeah |
| 07:43 | <MikeSmith> | problem is, I don't know how to describe that condition |
| 07:43 | <MikeSmith> | the condition is basically, if it ends up looking like one of the doctypes in the list below that, then true |
| 07:43 | <MikeSmith> | otherwise false |
| 07:44 | <zcorpan> | it's a bit confusing |
| 07:44 | <zcorpan> | i think Hixie's text is less confusing, at least to me :) |
| 07:45 | <MikeSmith> | I may end up changing it to that |
| 07:46 | <MikeSmith> | I wanted to first try to see if there were any way to make it any clearer |
| 07:46 | <zcorpan> | MikeSmith: maybe you could have two tables, one for html which has optional SI, and one for xhtml which doesn't |
| 07:46 | <MikeSmith> | hmm, yeah |
| 07:46 | <MikeSmith> | that would be slightly better at least |
| 07:47 | <MikeSmith> | zcorpan: I also changed the description of what a conformant XML document is (the other error you reported) |
| 07:48 | <hsivonen> | I wonder if a MacBook can be fooled to switch its VNC resolution to something higher than the res of the built-in screen without connecting anything to the mini-DVI port |
| 07:48 | <MikeSmith> | http://www.w3.org/TR/html-markup/documents.html#syntax-document-xml |
| 07:48 | <othermaciej> | annevk: looks like it's about 50/50 between what it is and what it isn't |
| 07:49 | <othermaciej> | it is funny that the one document not produced by the HTML WG (and something of a surprise to us) has by far the longest description |
| 07:49 | <MikeSmith> | zcorpan: I am planning to make a couple more tweaks based on the other message you sent, about "this specification does not define any additional syntax-level requirements" |
| 07:51 | <annevk> | othermaciej, the W3C clearly hates HTML5 much like it hates babies |
| 07:52 | <othermaciej> | annevk: I thought that was Adobe |
| 07:52 | <annevk> | oh, they're not the same? |
| 07:52 | <othermaciej> | heh |
| 07:52 | <othermaciej> | seriously though, HTML5 plus three other documents get less total space in the description than any other drafts |
| 07:52 | zcorpan | wonders what he has to do to see the browser choice screen |
| 07:54 | <zcorpan> | http://www.w3.org/TR/2010/WD-html-bidi-20100304/#br-as-separator - wonder if that matches what html5 says |
| 07:55 | <annevk> | zcorpan, visit http://www.browserchoice.eu/ ? |
| 07:56 | <othermaciej> | I dunno, it's funny that they used HTML4 as their baseline reference |
| 07:56 | <annevk> | zcorpan, I think dbaron mentioned that HTML5 should say something about that (iirc) |
| 07:56 | <annevk> | zcorpan, not sure if it was followed through |
| 07:56 | <zcorpan> | annevk: i thought i'd get something automatically by setting ie to default browser and doing a windows update |
| 07:57 | <annevk> | zcorpan, oh like that |
| 07:57 | <annevk> | zcorpan, dunno |
| 07:58 | <zcorpan> | how many of the non-top-5 browsers in the screen use trident? |
| 07:58 | <paul_irish> | maxthon does |
| 07:59 | <zcorpan> | yes, and flock doesn't, what about the others? |
| 07:59 | <zcorpan> | k-meleon also doesn't |
| 07:59 | <zcorpan> | avant is trident |
| 08:00 | <zcorpan> | looks like sleipnir is a trident browser |
| 08:01 | <zcorpan> | flashpeak too |
| 08:01 | <paul_irish> | and maybe AOL? |
| 08:02 | <othermaciej> | there are several trident browsers, according to the news |
| 08:03 | <othermaciej> | did they fix the randomization on that page yet? |
| 08:04 | <zcorpan> | trident: 6, gecko: 3, webkit: 2, presto: 1 |
| 08:05 | <othermaciej> | zcorpan: looks like it doesn't match HTML5 - "A br element does not separate paragraphs for the purposes of the Unicode bidirectional algorithm. [BIDI]" http://dev.w3.org/html5/spec/Overview.html#the-br-element |
| 08:08 | annevk | wonders what browsers implement now |
| 08:08 | <othermaciej> | according to the bidi requirements draft, WebKit and IE do not implement the HTML5/HTML4 rule but instead do treat <br> as a paragraph break for bidi algorithm purposes |
| 08:09 | <othermaciej> | the draft also says this behavior is better (I think) |
| 08:11 | <annevk> | I see |
| 08:12 | <annevk> | I wonder why they want through the trouble of publishing a separate document rather than sending feedback on HTML5... |
| 08:17 | <othermaciej> | I dunno |
| 08:18 | <othermaciej> | this could be, like, 10-20 bug reports |
| 08:18 | <othermaciej> | most of which would be uncontroversial except perhaps trivial syntax details |
| 08:19 | <othermaciej> | (for example the proposed bdi attribute should be a standard html-style boolean attribute; shouldn't use attributes with underscore; etc) |
| 08:20 | <annevk> | some of the guidelines should not be in HTML I think, e.g. the bit about <pre> in http://www.w3.org/TR/2010/WD-html-bidi-20100304/#newline-as-separator seems like it should be in CSS |
| 08:24 | <MikeSmith> | zcorpan: how's the following sound? |
| 08:24 | <MikeSmith> | [[ |
| 08:25 | <MikeSmith> | A conformant document in the XML syntax must be a well-formed XML document, as defined in the XML specification [XML], and its root element, as defined in the XML specification [XML], must be an html element. |
| 08:25 | <MikeSmith> | ]] |
| 08:25 | <zcorpan> | MikeSmith: the next paragraph seems to repeat the well-formedness requirement |
| 08:26 | <roc> | 'bdi' is a terrible name |
| 08:26 | <zcorpan> | MikeSmith: how about saying it must be namespace-well-formed, once? |
| 08:26 | <roc> | some of these issues were discussed briefly on public-html ages ago |
| 08:26 | <MikeSmith> | zcorpan: OK |
| 08:27 | <MikeSmith> | roc: yeah |
| 08:27 | <MikeSmith> | I recall now Richard mentioning the <br> issue to me a while back |
| 08:35 | <MikeSmith> | zcorpan: |
| 08:35 | <MikeSmith> | "A conformant document in the XML syntax must be a namespace-well-formed XML document, as defined in the XML specification [XML] and Namespaces in XML 1.0 specification [XMLNS], and its root element must be an html element." |
| 08:35 | <MikeSmith> | and I turned the "Documents in the XML syntax must not make use of any features of the HTML syntax that do not follow XML well-formedness constraints (for example, documents in the XML syntax must not use unquoted attribute value syntax and must not omit tags). |
| 08:35 | <MikeSmith> | " part into a note |
| 08:37 | <zcorpan> | s/as defined in the XML specification [XML] and // |
| 08:38 | <MikeSmith> | OK |
| 08:38 | <zcorpan> | isn't the root element required by the content models, btw? |
| 08:41 | <MikeSmith> | zcorpan: it's otherwise not explicitly stated anywhere for the XML/XHTML document case |
| 08:41 | <MikeSmith> | it is stated explicitly for the text/html case - |
| 08:41 | <MikeSmith> | file:///opt/workspace/html5/markup/documents.html#conformant-documents |
| 08:42 | <MikeSmith> | oops |
| 08:43 | <MikeSmith> | http://dev.w3.org/html5/markup/documents.html#syntax-document-html |
| 08:43 | zcorpan | gets a twitter account |
| 08:48 | <MikeSmith> | hsivonen: "operative encoding" |
| 08:51 | <Philip`> | Hixie: The analyse.cgi stuff is intentionally broken since I moved to a new server and don't really want to bother setting up all the databases and everything |
| 08:51 | <Philip`> | It's fixable but it depends on whether anybody cares enough |
| 08:53 | annevk | follows zcorpan |
| 08:57 | <Hixie> | Philip`: k |
| 10:43 | <mut> | # |
| 10:43 | <jgraham> | ! |
| 10:43 | <annevk> | ~ |
| 10:44 | <mut> | $(':(').show(); |
| 10:44 | <jgraham> | ‽ |
| 10:44 | <annevk> | you usually hide sad faces |
| 10:45 | <jgraham> | Only if you're a clown |
| 10:45 | <gsnedders> | jgraham: It's a Javascript library, duh. $ is awesome. :P |
| 10:45 | <gsnedders> | Normally my eyes are _above_ my mouth. |
| 10:46 | jgraham | is horribly confused |
| 10:46 | gsnedders | is good at making jgraham horribly confused |
| 10:46 | <jgraham> | You are quite the confusing person |
| 10:47 | <mut> | im about to get trout slappy |
| 10:47 | <beowulf> | i haven't been slapped by a wet trout in years |
| 10:47 | <mut> | "This program has stopped responding" <----- :( |
| 10:47 | <jgraham> | Fun game: read ꦕꦫꦏꦤ꧀ |
| 10:48 | <beowulf> | does the internet still have such things? |
| 10:48 | <jgraham> | (gsnedders isn't allowed to play) |
| 10:48 | <mut> | would you like to lose all of your work |
| 10:48 | <gsnedders> | Spoil sport. |
| 10:48 | mut | troutslaps beowulf |
| 10:48 | <mut> | ;) |
| 10:48 | workmad3 | slaps mut around a bit with a wet trout |
| 10:48 | <mut> | haha |
| 10:48 | <gsnedders> | jgraham is such a mean guy. |
| 10:48 | <mut> | ewwww moist |
| 10:48 | <workmad3> | you need to get a troutslap right :P |
| 10:49 | <mut> | ah im on a java applet, no troutslaps |
| 10:49 | <workmad3> | well, I just type em in manuallp |
| 10:49 | <workmad3> | *manually |
| 10:49 | jgraham | feels his game may have been lost in YAMPITS |
| 10:49 | <workmad3> | but I think mIRC used to have a /slap command to do it for you :) |
| 10:49 | <mut> | show a man a troutslap button, he will troutslap for a day, teach a man /me, and he will troutslap for life |
| 10:50 | <beowulf> | lol |
| 10:50 | <gsnedders> | jgraham: That's what you get for posting arbitary Unicode codepoints |
| 10:51 | <jgraham> | They're not arbitay. That's the point :) |
| 10:51 | <gsnedders> | Well, seeming I'm forbidden from playing, you could probably guess I know what they are. |
| 10:52 | <annevk> | jgraham, for that I would have to be funny |
| 10:52 | <beowulf> | jgraham: ok, explain the game, i didn't get to kill or blow anyone up |
| 10:53 | <Philip`> | Google is worringly fast |
| 10:53 | <jgraham> | annevk: Now you confused me |
| 10:53 | <gsnedders> | Go annevk! |
| 10:53 | <Philip`> | I searched for jgraham's string and it found the IRC logs from here |
| 10:54 | <jgraham> | That's how Hixie reads irc |
| 10:56 | annevk | waits for jgraham to implode |
| 10:57 | mut | sidesteps away from jgraham |
| 10:58 | annevk | advices people with nicks starting with a "j" to invoke /nick |
| 10:59 | <zcorpan> | jgraham: is it carakan? |
| 10:59 | <gsnedders> | zcorpan: yes. |
| 10:59 | <zcorpan> | yay |
| 11:01 | <annevk> | MessagePort.onmessage is somewhat ugly |
| 11:01 | <annevk> | but I guess it works |
| 11:03 | <jgraham> | zcorpan: Specifically is is the word "carakan" spelt out in Carakan symbols |
| 11:05 | <jgraham> | Sadly I can't find any fonts with support for it |
| 11:09 | <Philip`> | jgraham: Could probably make one easily by renumbering the font in http://hanacaraka.fateback.com/dok&down.htm |
| 11:12 | <jgraham> | Intersting |
| 11:14 | <Philip`> | Not really, because there's no reason to ever read or write anything in that language |
| 11:16 | <jgraham> | Sure there is. Just 30 minutes ago it would have helped you understand this very channel |
| 11:30 | <gsnedders> | Philip`: That's what I thought yesterday. |
| 11:30 | <gsnedders> | Philip`: If you're in school in Java it's useful |
| 11:40 | <virtuelv> | Opera's IRC client has /slap which slaps you around a bit with a wet trout |
| 11:41 | <virtuelv> | Aren't google going to start realtime indexing of results, soon? |
| 11:41 | <virtuelv> | as in you can use PubSubHubbub |
| 11:42 | <jgraham> | Yes |
| 11:44 | zcorpan | slaps a large trout around a bit with virtuelv |
| 11:45 | jgraham | thinks he has accidentially discovered that there are two bands called Camera Obscura |
| 11:45 | <gsnedders> | By Spotify being confused |
| 11:45 | <gsnedders> | ? |
| 11:45 | <zcorpan> | virtuelv: it's large, not wet |
| 11:45 | <jgraham> | gsnedders: Yes |
| 11:45 | <gsnedders> | I see that too often |
| 11:46 | <jgraham> | It's a little jarring when you go from melodic Scottish pop to... something |
| 11:46 | <jgraham> | s/pop/indie pop/ or something |
| 11:49 | <jgraham> | Ah "post hardcore" apparentley |
| 11:49 | <jgraham> | At least according to wikipedia |
| 11:49 | <gsnedders> | Oh dear. |
| 11:49 | <jgraham> | Oh dear isn't even close |
| 12:28 | <mut> | haha sweet, my canvas page works on my android phone! :) |
| 14:37 | <Lachy> | it just seems like the DE proponents won't be happy with anything less than XML-style namespaces, despite having so many extensibility points in HTML5 already. |
| 14:37 | <zcorpan> | http://chikuyonok.ru/ambilight/ |
| 14:38 | <Lachy> | nice. |
| 14:40 | <gsnedders> | Lachy: There's a lot of specs already written that are designed for the current namespace structure within the DOM, and none of the HTML5 extensbility points allow me to put, e.g., dc into HTML5 |
| 14:40 | <LoneStar99> | Morning all! |
| 14:41 | <Lachy> | gsnedders, so? I consider that a feature, not a bug |
| 14:41 | <gsnedders> | Lachy: But what do I do when I want to add DC to my markup? |
| 14:41 | <LoneStar99> | Can anything be drawn on canvas which does not use " beginPath(); "? |
| 14:42 | <gsnedders> | Lachy: So that I can take my HTML document and pass the DOM into my existing toolchain? |
| 14:42 | <Lachy> | what do you want to add DC markup for in the first place, which isn't catered for by existing metadata elements in HTML, or which couldn't be addressed through microformats or microdata? |
| 14:44 | <gsnedders> | Lachy: The summary of a section cannot be fetched programmatically; µf and µdata both require forks in my codebase at a DOM level for handling HTML differently to generic XML. |
| 14:44 | <Philip`> | LoneStar99: Not sure what you mean |
| 14:45 | <Philip`> | LoneStar99: You ought to use beginPath every time you start drawing a new path, but things like drawImage and fillRect don't use paths |
| 14:45 | <Lachy> | gsnedders, are you giving a serious use case here and actually arguing that we should support namespaces in HTML, or are you just being hypothetical? |
| 14:45 | <LoneStar99> | in other words, can a drawing "sketching" be done without using "beginPath();" "moveTo" and "lineTo" |
| 14:45 | <Philip`> | LoneStar99: You could do it by just drawing images or drawing lots of little rectangles, but that'd be a silly way to draw lines compared to using lineTo :-) |
| 14:45 | <gsnedders> | Lachy: I'm trying to show you there are serious use cases here, and that people do have codepaths to deal with these things and supporting the HTML 5 way of doing things leads to little gain but a lot of additional complexity. |
| 14:47 | <Lachy> | sure, but the additional complexity of trying to support namespaces in HTML, compared with the little gained by supported those largely obscure and insignificant cases in practice, proves it's just not worth it |
| 14:47 | <gsnedders> | No, the additional complexity _depends_ upon the serialization chosen. |
| 14:48 | <Lachy> | then don't choose HTML if you depend on XML-specific features |
| 14:48 | <jgraham> | Requiring namespaces introduces a lot of complexity for authors |
| 14:48 | <jgraham> | By comparison I have rather little sympathy for implementors |
| 14:48 | <gsnedders> | Lachy: Feeds online are not an obscure or insignificant case, and currently require duplicated data |
| 14:49 | <gsnedders> | jgraham: Indeed, I think there are plenty of ways to cope with namespaces without additional complexity for authors in the pure-HTML case. |
| 14:49 | <Lachy> | gsnedders, wtf do feeds have to do with supporting namespaces in the HTML serialisation? |
| 14:49 | <jgraham> | e.g. not having namespaces? |
| 14:49 | <jgraham> | (and I also don't get the relevance of feeds) |
| 14:50 | <LoneStar99> | Philip`: how would I be able to drawing a sketching "drawing" tons of rectangles? |
| 14:50 | <LoneStar99> | with |
| 14:50 | <gsnedders> | Lachy: Why do I have to post both an HTML and Atom edition of my blog for feed readers? Why can feed readers not programmatically read content from the HTML? |
| 14:50 | <Philip`> | gsnedders: Because HTML parsers are rubbish and buggy |
| 14:51 | <Philip`> | or slow |
| 14:52 | <jgraham> | gsnedders: They can. There is no problem here |
| 14:52 | <gsnedders> | Philip`: All OS X feed readers have a decent HTML parser available at an OS level; I presume the same is true on Windows with mshtml.ddl |
| 14:52 | <gsnedders> | jgraham: But they cannot extract all the semantics they can get with Atom |
| 14:53 | <jgraham> | But DE wouldn't magically make that happen because in practice no one would use the extensions anyway |
| 14:53 | <Philip`> | gsnedders: Does it provide an actual standalone HTML parser than gives you back a DOM and is just as easy to use as XML parser libraries? |
| 14:53 | <jgraham> | An author that cares can already use the HTML->Atom conversion algorithm to get almost everything |
| 14:53 | <LoneStar99> | Philip': issue I have with "beginPath();" is that my code used to work without a "moveTo();" the code now, needs a " moveTo();" and notice on drawings, line are automatically connected when they should not be connected |
| 14:54 | <gsnedders> | Philip`: My memory from when I looked at it before was that it wasn't very hard. |
| 15:05 | <Lachy> | gsnedders, there have been various attempts at doing that (hAtom, the "extract an Atom feed" algorithm in HTML5), but so far it seems the market isn't really interested in pushing those kinds of techniques. Besides having a CMS generate an Atom feed automatically is a solved problem that authors just don't have to worry about. |
| 15:09 | <gsnedders> | So duplicated data isn't a problem when we have tools? |
| 15:12 | <Lachy> | if it wasn't already a solved problem, I'd care more about the duplication |
| 15:21 | <Philip`> | gsnedders: Data isn't duplicated - there's a single copy of each blog post in the database |
| 15:21 | <Philip`> | and so you can avoid all the problems like duplicates getting out of sync with each other |
| 15:40 | <mut> | anyone know of a good canvas tutorial? |
| 15:45 | <LoneStar99> | mut: what u trying to do with canvas? |
| 15:46 | <miketaylr> | mut: http://dev.opera.com/articles/view/html-5-canvas-the-basics/ && https://developer.mozilla.org/en/Canvas_tutorial are pretty good |
| 15:46 | <mut> | im drawing staircases in canvas |
| 15:47 | <mut> | but just thinking of some other things to do, which are currently beyond me :P |
| 15:47 | <LoneStar99> | yeah, try the above links, those are the ones I have uded |
| 15:47 | <LoneStar99> | used |
| 15:47 | <mut> | thanks mike btw :) |
| 15:48 | <mut> | yea im sure I can do what i was planning, but I have had a few more ideas for usablility (zoom and drag functionality ala googlemaps) |
| 16:13 | <TabAtkins> | Oh man, only 5 days before my lightcone envelopes HR222. |
| 16:15 | <Philip`> | http://thomas.loc.gov/cgi-bin/bdquery/z?d109:h.r.00222: - "To prohibit the expenditure of Federal funds to conduct or support research on the cloning of humans, and to express the sense of the Congress that other countries should establish substantially equivalent restrictions." |
| 16:16 | <Philip`> | TabAtkins: ^ That HR222? |
| 16:16 | <TabAtkins> | No, look for the first one involving stars. |
| 16:17 | <Philip`> | Oh |
| 16:18 | <Philip`> | Happy almost about 24.3rd birthday, then |
| 16:20 | <Dashiva> | http://twitter.com/heycam/status/10003477514 |
| 16:20 | <Dashiva> | Maybe I should tell heycam how buses in Norway don't play radio anymore because of inspectors from the local RIAA equivalent |
| 16:47 | <mut> | hmm opera dosent seem to handle save() and restore() when combined with scale()...well not the same as firefox or ie +excanvas |
| 16:48 | <Philip`> | The problem might be that it applies the transformation when you call stroke/fill, instead of applying it when you're adding points to the path |
| 16:48 | <Philip`> | (so if you change the transformation between constructing the path and stroking/filling then it'll break) |
| 17:11 | <AryehGregor> | Where are the W3C Editor's Drafts? |
| 17:11 | <AryehGregor> | They seem to be cunningly hidden. |
| 17:11 | <AryehGregor> | Even the link on the HTMLWG page to the latest editor's draft links to a WD. |
| 17:12 | <Philip`> | http://dev.w3.org/html5/ ? |
| 17:13 | <AryehGregor> | Oh, wait, I get it. |
| 17:13 | <Philip`> | I think the ones like http://dev.w3.org/html5/spec/ just haven't been changed back to the ED colours yet |
| 17:13 | <AryehGregor> | The latest editor's drafts are actually WDs right now? |
| 17:13 | <AryehGregor> | Or something. |
| 17:17 | <Philip`> | AryehGregor: Yes |
| 17:17 | <Philip`> | They were changed in e.g. http://dev.w3.org/cvsweb/html5/spec/index.html.diff?r1=1.807&r2=1.808 |
| 17:18 | <Philip`> | ("front matter updates for publication this week", 2.5 weeks ago) |
| 17:18 | <Philip`> | and will presumably be changed back soon |
| 17:26 | <LoneStar99> | Philip': you know of a tutorial, of how to draw a lines with tons of squares, like you mentioned before? |
| 17:27 | <Philip`> | LoneStar99: I expect there's lots of things talking about how to implement line rasterization algorithms, but it seems like a terrible idea :-) |
| 17:28 | <LoneStar99> | Philip: it does, but i know of an app, that does it and works really well... |
| 17:29 | <annevk> | omg http://www.w3.org/Bugs/Public/show_bug.cgi?id=9197 |
| 17:29 | <annevk> | evil |
| 17:29 | <Philip`> | Using beginPath/moveTo/lineTo is the right way to draw lines, and it should be able to do everything you need, so if your code has bugs then you just need to fix those bugs |
| 17:30 | <LoneStar99> | Philip`: the thing it might not be my app, it might be the OS and need to find a work around... |
| 17:33 | <Philip`> | That seems unlikely |
| 17:33 | <LoneStar99> | tried everything... it just seems that my app randomly begins to connect lines and i think it has to do with "context.moveTo(this.previous_x, this.previous_y);" but it need the line to work... |
| 17:35 | <Philip`> | It sounds like you just have a bug somewhere, but I don't know what your code is doing so I have no idea where it might be wrong |
| 17:38 | <LoneStar99> | Philip' http://pastebin.com/uNxvLdL4 |
| 17:43 | <Philip`> | LoneStar99: You need to set previous_x and previous_y in mousedown, otherwise they'll be left as whatever they were in the last mousemove even if you released the mouse, I guess |
| 17:46 | <LoneStar99> | Philip': have tried that and it still begins to screw up after awhile |
| 17:47 | <TabAtkins> | Tried logging the events and the values they see for previous_x and previous_y? |
| 17:48 | <Philip`> | LoneStar99: By the way, why is the drawImage line there? |
| 17:48 | <Philip`> | Also, why is the stroke() not at the end of mousemove? |
| 17:50 | <tantek> | Hixie, does microdata itemref permit referencing an id of an ancestor? |
| 17:50 | <tantek> | e.g. <span itemscope id=a><span itemprop=turtle itemscope id=b itemref=a></span></span> |
| 17:51 | <TabAtkins> | It might run afoul of the cycle-breaking algo, but there's no explicit restriction against such a thing. |
| 17:51 | <TabAtkins> | I haven't kept up with the cycle algo, though. |
| 17:51 | <TabAtkins> | foolip would have more info, I think. |
| 17:51 | <tantek> | TabAtkins - what cycle breaking algo? none specified here: http://www.w3.org/TR/microdata/#attr-itemref |
| 17:52 | <tantek> | it just says "a list of additional elements to crawl to find the name-value pairs of the item." |
| 17:53 | <tantek> | which in the example above, gives you infinitely nested items, each with a turtle property whose value is the next nested item with a turtle property etc. |
| 17:53 | <tantek> | turtles all the way down ;) |
| 17:53 | <tantek> | when do you stop crawling? |
| 17:53 | <LoneStar99> | Philip': i makes things look anti-aliasing |
| 17:54 | <TabAtkins> | tantek: http://www.whatwg.org/specs/web-apps/current-work/multipage/microdata.html#associating-names-with-items |
| 17:54 | <TabAtkins> | Dunno if this is the latest version of the cycle-breaker. You'd have to ask foolip, or get Hixie. |
| 17:55 | <tantek> | ah - ok: If root is in memory, then the algorithm fails; abort these steps. |
| 17:55 | <tantek> | therefore microdata effectively makes referencing an id of an ancestor a NOP |
| 17:56 | <TabAtkins> | If the ancestor is already part of the item, yes. |
| 17:56 | <TabAtkins> | If it's an ancestor in the DOM, higher up than the start of the item, though, then should be okay. |
| 17:56 | <tantek> | sure |
| 17:56 | <tantek> | microformats include-pattern has the same restriction |
| 17:57 | <tantek> | but I didn't see it in the W3C microdata draft |
| 17:57 | <tantek> | so I wondered if Ian had somehow permitted itemref inclusion of ancestors |
| 17:57 | <tantek> | apparently not |
| 17:57 | <TabAtkins> | Is that algo in the w3c draft, or are the two versions differeing significantly? |
| 17:58 | <LoneStar99> | Philip': Maybe not, the drawImage stuff does nothing from i can see |
| 17:59 | <tantek> | TabAtkins - the w3c draft says the same: http://www.w3.org/TR/microdata/#associating-names-with-items |
| 17:59 | <TabAtkins> | kk. |
| 17:59 | <tantek> | I just didn't find it because it doesn't reference itemref explicitly |
| 17:59 | <tantek> | itemref on an ancestor being a NOP is a side effect you have to figure out from the algo |
| 17:59 | <tantek> | rather than being explicit |
| 17:59 | <TabAtkins> | Yes. |
| 17:59 | <tantek> | which is going to confuse authors |
| 18:00 | <tantek> | when they ask why doesn't this work |
| 18:00 | <TabAtkins> | The simple and obvious answer would be "because that would make an infinite loop, don't do that". |
| 18:00 | <TabAtkins> | You don't need to point them at the algo for that. |
| 18:02 | <tantek> | much better to describe it up front IMHO |
| 18:02 | <tantek> | e.g. |
| 18:02 | <tantek> | http://microformats.org/wiki/include-pattern#in_general |
| 18:02 | <tantek> | To prevent infinite loops, if a class="include" refers to itself or to an ancestor in the parse tree, then it is ignored and has no effect on the parser. |
| 18:02 | <LoneStar99> | are there any webOS developers here? |
| 18:03 | <TabAtkins> | tantek: But the situation is more subtle with microdata, because it permits reference to ancestors; it merely prevents such a reference from causing an infinite loop. |
| 18:03 | <tantek> | LoneStar99 try #webkit |
| 18:04 | <tantek> | TabAtkins - not true - the ancestor is already in memory from the parse in progress |
| 18:04 | <tantek> | thus the reference to ancestor immediately fails in step 1 |
| 18:04 | <tantek> | assuming you're parsing microdata from the root of document downward |
| 18:05 | <tantek> | which all the "converting to ..." (insert format here) algorithms do |
| 18:05 | <LoneStar99> | tantek: there is a webos channel, but wanted to see if someone could see the issues i see with my code... but anyways time for lunch! thanks for all the help be back later |
| 18:05 | <tantek> | LoneStar99 - good to know! |
| 18:08 | <tantek> | btw - given that itemref causes an actual crawl of the targets of the itemref it is not really a "ref" (like href), it is more like a "src" (like <img src>) |
| 18:08 | <tantek> | itemsrc would more accurately convey what itemref does |
| 18:08 | <TabAtkins> | tantek: I believe that that algo starts by assuming the root is an item, that is, it has @itemscope. |
| 18:09 | <tantek> | TabAtkins - sure - but that makes no difference to the breaking of recursion |
| 18:09 | <tantek> | or ancestral reference |
| 18:09 | <tantek> | alternatively, iteminclude would also better convey what itemref does |
| 18:10 | <TabAtkins> | <div id=foo itemprop=foo>foo <div itemscope itemref=foo></div></div> would affect it, wouldn't it? |
| 18:13 | <tantek> | TabAtkins - yes - not a particularly useful case though |
| 18:13 | <tantek> | as typically you get a need in nested objects to refer to (include) a common parent object |
| 18:13 | <TabAtkins> | No, but that's the reason why Ian often forgoes prose descriptions of algos - they don't hit all cases. |
| 18:14 | <tantek> | sure, but then that makes it harder for authors |
| 18:14 | <tantek> | authors don't typically "get" the procedural descriptions of the algos |
| 18:14 | <TabAtkins> | Indeed. |
| 18:15 | <tantek> | and one of the claims of microdata is that it is "easier" for authors |
| 18:15 | <tantek> | thus I would say that claim is refuted by this |
| 18:15 | <tantek> | TabAtkins - if you want to see a discussion of a real world example of nested objects referring to a parent object, /join #microformats |
| 18:16 | <tantek> | we're discussing hProduct + hReview there now |
| 18:16 | <TabAtkins> | Oh, I don't doubt the use-case. I use circularly-nested data structures all the time. |
| 18:16 | <tantek> | all the time? |
| 18:16 | <tantek> | or are you being facetious ;) |
| 18:16 | <TabAtkins> | No, I really do use them a lot. |
| 18:17 | <TabAtkins> | It's very useful to embed a reference to the parent in a child. |
| 18:18 | <tantek> | a reference yes. but an include no. and itemref is an include, not a reference. hence my point about terminology. |
| 18:18 | <TabAtkins> | That's valid. |
| 18:19 | <TabAtkins> | Then suggest some accurate prose, or suggest a change to the algo to reflect the simpler prose. |
| 18:19 | <TabAtkins> | Such as disallowing all ancestor references. |
| 18:19 | TabAtkins | goes to lunch. |
| 18:20 | <JonathanNeal_oww> | Hey all. |
| 18:20 | <JonathanNeal> | I'm trying to come up with practical naming conventions for the IDs and Classnames in my layout template. |
| 18:21 | <JonathanNeal> | I was trying to follow the ARIA roles list to name most of my elements. |
| 18:58 | <Philip`> | JonathanNeal: I suggest using boys' names for classes and girls' names for IDs |
| 18:58 | <JonathanNeal> | Ha. |
| 18:58 | <Philip`> | Helps to keep them easily distinguishable |
| 18:58 | <JonathanNeal> | I found an ARIA name. |
| 18:58 | <JonathanNeal> | role="main" so I named it main. |
| 19:07 | <Hixie> | tantek: please send microdata feedback to the list or use the review comments tool in the whatwg version to file a bug so i don't lose it -- thanks |
| 19:18 | <mpilgrim> | I don't really understand the purpose of the "fallback" section in a cache manifest file |
| 19:19 | <mpilgrim> | i see the use case for the "explicit" section (stuff you need offline) |
| 19:19 | <mpilgrim> | and i see the use case for the "online whitelist" section (stuff like tracking CGI scripts that don't make sense offline) |
| 19:20 | <mpilgrim> | there's only one example in the spec of using a "fallback" section, and i don't fully understand it |
| 19:20 | <mpilgrim> | scratch that, i don't understand it at all |
| 19:22 | <Hixie> | imagine a site like flickr |
| 19:22 | <Hixie> | which has one page per image |
| 19:23 | <Hixie> | now imagine the user, while offline, adds a bunch of images |
| 19:23 | <Hixie> | and follows links to those images |
| 19:23 | <Hixie> | if we didn't have the fallback section, those links would be dead |
| 19:24 | <mpilgrim> | ... |
| 19:25 | <mpilgrim> | ok, so offline flickr |
| 19:26 | <mpilgrim> | suppose http://www.flickr.com/photos/f8dy/ links to a manifest file |
| 19:27 | <Hixie> | crap meeting gotta go |
| 19:28 | <mpilgrim> | http://www.flickr.com/photos/f8dy/327774195/ (and, say, 99 other recent photo pages) is listed in the explicit section |
| 19:28 | <mpilgrim> | along with http://farm1.static.flickr.com/142/327774195_2d2b67753e.jpg which is the image on that page |
| 19:31 | <tantek> | Hixie, will do. For reference, my feedback items are described here: http://krijnhoetmer.nl/irc-logs/whatwg/20100305#l-781 I'll clean up that text and send it to public-html. |
| 19:32 | tantek | was trying to figure out what about that photo of towels over mpilgrim's feed made it "explicit". |
| 19:32 | <mpilgrim> | haha |
| 19:33 | <mpilgrim> | pooh porn |
| 19:33 | <mpilgrim> | rule 34 predicts its existence |
| 19:36 | <mpilgrim> | a fun fact about the cache manifest: if there are no explicit section headers, all the listed resources are implicitly in the "explicit" section |
| 19:37 | <mpilgrim> | i tried to explain this with a straight face for my book, but i failed |
| 20:08 | <knowtheory> | Hey othermaciej :) |
| 20:08 | <othermaciej> | hello |
| 20:11 | <knowtheory> | are CSS animations incorporated into a component of HTML5 or into a new CSS standard? |
| 20:11 | <knowtheory> | I'm trying to figure out the degree of complexity and organizational capaobilities of CSS animations (basically, how one would work w/ 'em for complex examples) |
| 20:24 | <TabAtkins> | knowtheory: CSS Animations are (unsurprisingly ^_^) a CSS Module. |
| 20:24 | <knowtheory> | TabAtkins: in HTML5 then? :D |
| 20:24 | <TabAtkins> | Haha. |
| 20:25 | <knowtheory> | hunh, exciting, http://www.w3.org/TR/css3-ui/ just crashed a bunch of tabs in chrome |
| 20:26 | <knowtheory> | yeah freaky, consistent crash on that page. |
| 20:26 | <knowtheory> | safari renders it fine though |
| 20:26 | <TabAtkins> | Huh, no crash for me. Windows, Chrome 4. |
| 20:27 | <knowtheory> | yeah osx 5.0.307.11 b |
| 20:27 | <miketaylr> | works fine in 5.0.345.0 (40651)/Mac |
| 20:27 | <miketaylr> | (well, chromium) |
| 20:27 | <knowtheory> | even more exciting! :) |
| 20:37 | <knowtheory> | ah there we go: http://dev.w3.org/csswg/css3-animations/ |
| 20:42 | <TabAtkins> | knowtheory: Ah man, I'm sorry, I would have linked you to it immediately if I knew you were looking for it. |
| 20:43 | <knowtheory> | TabAtkins: no worries! |
| 20:59 | <tantek> | I did encode actual example uses of CSS3 UI properties in the spec itself - I wonder if that may be causing the crash - certainly was not intentional. |
| 21:05 | <knowtheory> | tantek: it's all your fault! |
| 21:05 | <knowtheory> | since someone mentioned that chromium wasn't seeing that behavior |
| 21:05 | <knowtheory> | i figure it's just a bug in this version of chrome |
| 21:07 | <miketaylr> | knowtheory: heh, i think you're right: http://miketaylr.com/post/bcbf19c8.png |
| 21:07 | <miketaylr> | i crash using the same version of chrome, but am fine in chromium |
| 21:08 | <knowtheory> | miketaylr: i filed a bug on the chromium tracker :) |
| 21:09 | <miketaylr> | cool |
| 21:11 | <knowtheory> | miketaylr: http://code.google.com/p/chromium/issues/detail?id=37546 |
| 21:11 | <knowtheory> | (also for you tantek if you are interested) |
| 21:16 | <knowtheory> | man, that was fast, the bug i reported has already been triaged and assigned |
| 21:25 | <dglazkov> | knowtheory: speed is what we do :) |
| 21:25 | <dglazkov> | that didn't come out right |
| 21:26 | <knowtheory> | haha :D |
| 21:27 | <mpilgrim> | woohoo, i have a working offline web application |
| 21:28 | <mpilgrim> | whose sole purpose is to display information on whether or not it is a working offline web application |
| 21:30 | <mpilgrim> | it works in 2 browsers! |
| 21:31 | <mpilgrim> | my work here is done |
| 21:38 | <othermaciej> | hawt |
| 21:38 | <othermaciej> | annevk: does html5-diffs have a bug component? |
| 21:52 | <mpilgrim> | no, scratch that |
| 21:52 | <mpilgrim> | i can't get it to register updates in firefox |
| 21:53 | <mpilgrim> | http://diveintohtml5.org/examples/clock.html |
| 21:53 | <mpilgrim> | manifest is at http://diveintohtml5.org/examples/clock.html5cache |
| 21:53 | <Hixie> | why "html5"cache? |
| 21:53 | <mpilgrim> | firefox properly downloads it and can use it offline |
| 21:54 | <mpilgrim> | but when i update the JS file with new code and update the manifest file with a version-specific comment, it never picks up the updated resources |
| 21:54 | <Hixie> | does it try to fetch the manifest when you load the page? |
| 21:54 | <mpilgrim> | curl -I confirms that the etag and last-modified headers of the manifest file are changing |
| 21:55 | <mpilgrim> | oh bloody hell, my expires headers are wrong |
| 21:55 | <othermaciej> | what's the expiration on the manifest? |
| 21:55 | <mpilgrim> | yeah, just got there |
| 21:55 | <mpilgrim> | hang on |
| 22:07 | <temoto> | Hello. I want to try html5lib instead of BeautifulSoup to parse arbitrary HTML found on the web. What i'm [currently] interested is only about finding <a> elements and taking their href attribute value. (as opposed to keeping Soup interface) So which is the fastest way to iterate all <a> elements? |
| 22:14 | <mpilgrim> | woohoo, http://diveintohtml5.org/examples/offline/clock.html works in 2 browsers! |
| 22:14 | <mpilgrim> | hixie: i don't know why i called it ".html5cache" |
| 22:15 | <mpilgrim> | i needed a unique file extension to set the content-type |
| 22:15 | <Hixie> | and you didn't like .manifest like the spec suggests? :-P |
| 22:16 | <mpilgrim> | i missed that |
| 22:16 | <mpilgrim> | that doesn't conflict with anything else in MIME-world? |
| 22:16 | <Hixie> | not to my knowledge |
| 22:17 | <mpilgrim> | ("MIME-world" reminds me of the underwater minus level in Super Mario Bros. -- hard to get into and impossible to win) |
| 22:24 | <mpilgrim> | while we're on the subject, do you have a suggestion for a MIME type for .ttf files? |
| 22:24 | <Hixie> | font/truetype? |
| 22:24 | <Hixie> | might need registering |
| 22:24 | <mpilgrim> | it's one of those MIME-world trick questions, i think |
| 22:25 | <mpilgrim> | AFAICT, there's nothing registered |
| 22:25 | <mpilgrim> | and all the experimental or vendor-specific alternatives are horrific |
| 22:25 | <Hixie> | ah. then you just volunteered to register it! |
| 22:25 | <mpilgrim> | apparently it doesn't matter because font-face-supporting browsers ignore the mime type |
| 22:26 | <Hixie> | yeah well |
| 22:26 | <mpilgrim> | but i got email complaints because my fonts were being served as text/plain, which is arguably the worst default ever |
| 22:26 | Hixie | wishes we could abandon MIME types for anything that can be unambiguously identified by the leading bytes |
| 22:26 | <mpilgrim> | and judging by the online discussions i've found on the topic, registering it is impossible |
| 22:26 | <Hixie> | why is registering it impossible? |
| 22:27 | <mpilgrim> | maybe not theoretically impossible |
| 22:27 | <mpilgrim> | but no one has managed it yet |
| 22:27 | <mpilgrim> | despite years of trying |
| 22:27 | <mpilgrim> | and no, i'm not volunteering to register it |
| 22:28 | <mpilgrim> | i don't need any more email from Those People |
| 22:30 | <Hixie> | <img alt="Stamp It" onerror="this.src='/icon.gif';" src="" align="left" border="0" height="32" width="32"> |
| 22:30 | <Hixie> | wtf |
| 22:30 | <mpilgrim> | hooray for alt text! |
| 22:30 | <mpilgrim> | does that ever work? |
| 22:30 | <mpilgrim> | (the onerror part, i mean) |
| 22:33 | <Hixie> | apparently |
| 22:33 | <Hixie> | src="" triggers onerror, which sets the icon, or something |
| 22:33 | <mpilgrim> | awesome |
| 22:34 | <mpilgrim> | so you were explaining the fallback section of the manifest |
| 22:34 | <mpilgrim> | using flickr as an example |
| 22:34 | <Hixie> | yes |
| 22:34 | <mpilgrim> | i don't understand it yet |
| 22:35 | <Hixie> | it lets you cache a small part of a site and have graceful fallback for the rest of the site while offline |
| 22:37 | <mpilgrim> | and by "graceful fallback" you mean "a page that says this page is not available offline" |
| 22:37 | <mpilgrim> | in lieu of a browser-level error message |
| 22:37 | <mpilgrim> | ? |
| 22:37 | <Hixie> | not necessarily; if you have the information locally to render the page, e.g. if bugzilla has a copy of the bug in its client-side database, you can just render the page locally |
| 22:38 | <temoto> | Hello. I want to try html5lib instead of BeautifulSoup to parse arbitrary HTML found on the web. What i'm [currently] interested is only about finding <a> elements and taking their href attribute value. (as opposed to keeping Soup interface) So which is the fastest way to iterate all <a> elements? |
| 22:39 | <mpilgrim> | that makes sense for bugzilla because each page is "show_bug.cgi?id=foo" and the only thing that changes is foo |
| 22:39 | <mpilgrim> | earlier you mentioned a hypothetical offline flickr |
| 22:39 | <mpilgrim> | flickr uses unique URLs for each photo page |
| 22:41 | <mpilgrim> | are you saying that you could set up a wildcard so that all unknown (uncached) URLs got handled by a master page like show_photo.cgi?id=foo |
| 22:41 | <Hixie> | i just meant an image site in general, dunno abotu flickr specifically |
| 22:41 | <mpilgrim> | ? |
| 22:41 | <Hixie> | that's what the fallback stuff does, yes |
| 22:42 | <mpilgrim> | i see |
| 22:43 | <mpilgrim> | that was the missing link |
| 22:43 | <mpilgrim> | i didn't know that fallback URLs could be wildcards |
| 22:43 | <mpilgrim> | now i see the bit about "fallback namespace" and "URLs as prefix match patterns" |
| 22:43 | <Hixie> | they work the same as the online whitelist, but with a fallback url |
| 22:44 | <mpilgrim> | can i declare a specific URL in the explicit section that would also match a fallback namespace? |
| 22:44 | <mpilgrim> | like |
| 22:44 | <mpilgrim> | CACHE MANIFEST |
| 22:44 | <mpilgrim> | CACHE: |
| 22:44 | <mpilgrim> | images/cat.jpg |
| 22:45 | <mpilgrim> | FALLBACK: |
| 22:45 | <mpilgrim> | images/ missing.jpg |
| 22:45 | <Hixie> | what do you mean by "can i" exactly? |
| 22:46 | <mpilgrim> | will the browser cache images/cat.jpg and use it when i say <img src="images/cat.jpg"> ? |
| 22:46 | <mpilgrim> | the URL matches a resource listed in the explicit section, and also a fallback namespace listed in the fallback section |
| 22:46 | <mpilgrim> | which one wins? |
| 22:46 | <Hixie> | yes |
| 22:46 | <Hixie> | If a resource is listed in the explicit section or as a fallback entry in the fallback section, the resource will always be taken from the cache, regardless of any other matching entries in the fallback namespaces or online whitelist namespaces. |
| 22:46 | <mpilgrim> | ok |
| 22:48 | <mpilgrim> | now here's another question |
| 22:48 | <mpilgrim> | i've been trying to set up all of diveintohtml5.org as an offline web app |
| 22:48 | <mpilgrim> | but i don't want to do it by default |
| 22:49 | <mpilgrim> | i want you to click a link "go offline" and go to a page that gives a progress bar |
| 22:49 | <mpilgrim> | (in firefox this page would also display the prompt "do you want this site to store local data" or some such) |
| 22:49 | <mpilgrim> | (in safari there is apparently no prompt) |
| 22:49 | <mpilgrim> | IOW, i don't want every page to point to the manifest file |
| 22:50 | <Hixie> | if a page doesn't point to the manifest, it can't be cached for offline use |
| 22:50 | <mpilgrim> | but i want the go_offline.html page to point to ta manifest file that includes all the other HTML pages (and their scripts, stylesheets, and images) |
| 22:50 | <Hixie> | each page that is to work offline has to opt in to working offlien |
| 22:51 | <mpilgrim> | so page A *must* point to a manifest file in order to be used offline? |
| 22:51 | <mpilgrim> | even if page B points to a manifest file that includes page A? |
| 22:52 | <Hixie> | yes |
| 22:52 | <mpilgrim> | even after the user visits page B and consents to storing local data? |
| 22:52 | <Hixie> | yes |
| 22:52 | <mpilgrim> | that sucks |
| 22:53 | <mpilgrim> | i don't want every page of my site popping up a scary infobar about "storing local data" |
| 22:53 | <Hixie> | your site isn't an application |
| 22:53 | <mpilgrim> | sure it is |
| 22:53 | <mpilgrim> | it's an interactive ebook |
| 22:53 | <mpilgrim> | it has interactive elements |
| 22:54 | <mpilgrim> | and scripts that give live results based on the current browser environment |
| 22:54 | <LoneStar99> | is Philip` around? |
| 22:54 | <Philip`> | Maybe |
| 22:54 | <Hixie> | mpilgrim: then just stick a manifest on each page |
| 22:54 | <Hixie> | mpilgrim: and let the user decide whether to go offline or not |
| 22:55 | <LoneStar99> | Philip` have a video with the issue at hand http://video.yahoo.com/watch/7098996/18475648 |
| 22:55 | <mpilgrim> | that user experience sucks |
| 22:55 | Philip` | can't watch videos currently |
| 22:55 | <Hixie> | the user experience is in the hands of the user agents, so get them to make it better |
| 22:56 | <LoneStar99> | Philip` ok will ping you later |
| 22:56 | <mpilgrim> | that's a shitty cop-out of an answer |
| 22:56 | <mpilgrim> | this was possible in gears, was it not? |
| 22:56 | <mpilgrim> | because everything was scripted, including the downloading of the manifest file, etc. |
| 22:56 | <Hixie> | no idea |
| 22:57 | <Hixie> | letting the user agent be in charge of the user experience is not a cop-out answer, it's the way the web is designed to work |
| 22:58 | <LoneStar99> | if anyone has exprience with canvas can you look at the following video and tell me what is going on: http://video.yahoo.com/watch/7098996/18475648 |
| 22:58 | <mpilgrim> | in html5, you can't script the downloading of the manifest file because you can't dynamically set the manifest attribute |
| 22:59 | <Hixie> | correct |
| 22:59 | <Hixie> | a page is either part of an application or it isn't |
| 22:59 | <mpilgrim> | i want a page to decide at runtime whether it is part of an application or it isn't |
| 23:00 | <Hixie> | can't be done |
| 23:00 | <Hixie> | well, you could have the server dynamically decide whether to set manifest="" or not, i guess |
| 23:00 | <Hixie> | but you'd have to reload the page or load it in an iframe or something to trigger teh load |
| 23:00 | <mpilgrim> | based on what? HTTP request headers? |
| 23:01 | <Hixie> | cookies, probably |
| 23:01 | <mpilgrim> | i suppose that would work |
| 23:01 | <Hixie> | it's not the intended use case though |
| 23:01 | <Hixie> | the intended use case is applications that are unambiguously applications |
| 23:02 | <mpilgrim> | MY SITE IS UNAMBIGUOUSLY AN APPLICATION |
| 23:02 | <mpilgrim> | it is ambiguously an offline application |
| 23:02 | <Hixie> | your site is apparently insecure about its application status, if it's not willing to always be an offline application :-) |
| 23:03 | <mpilgrim> | how are applications like gmail supposed to migrate to html5 if they can't offer an online application and an offline application at the same URL? |
| 23:03 | <Hixie> | what's the difference? |
| 23:04 | <Hixie> | between an "online" app and an "offline" app, i mean |
| 23:05 | <mpilgrim> | i navigate to http://mail.google.com/ in firefox |
| 23:05 | <mpilgrim> | i log in |
| 23:05 | <mpilgrim> | i click "settings" |
| 23:05 | <mpilgrim> | i click "offline" |
| 23:06 | <Hixie> | why? |
| 23:06 | <mpilgrim> | i click "enable offline access" |
| 23:06 | <Hixie> | here's how it should work: |
| 23:06 | <Hixie> | you navigate to http://mail.google.com/ in firefox |
| 23:06 | <Hixie> | you log in |
| 23:06 | <Hixie> | end of story |
| 23:06 | <mpilgrim> | it pops up a pseudo-dialog titled "install offline access for gmail" |
| 23:07 | <mpilgrim> | because not every user who logs into gmail wants, needs, or is operating in a safe browser environment to have offline access |
| 23:07 | <mpilgrim> | the pseudo-dialog i am reading right now says "This feature will download your email messages onto this computer. Please make sure it is not a public or shared computer." |
| 23:08 | <Hixie> | sure, actually downloading the mail might be something you want to do after checking a pref |
| 23:08 | <Hixie> | but that's just making the mail available offline, that's not making the _app_ available offline |
| 23:09 | <Hixie> | the data part has nothing to do with the cache manifest |
| 23:09 | <Hixie> | you can download that without making an "offline" app |
| 23:09 | <mpilgrim> | so your ideal user experience is signing up for gmail, logging in for the first time, and getting an infobar saying "this site is asking to store local data [allow] [deny] [learn more]" |
| 23:09 | <Hixie> | no |
| 23:09 | <Hixie> | there's no reason for the user agent to ask the user about caches |
| 23:09 | <Hixie> | it's basically no different than an HTTP cache |
| 23:10 | <mpilgrim> | but firefox does, demonstrably, ask the user for permission before storing any data in the applicationCache |
| 23:10 | <Hixie> | file a bug |
| 23:10 | <Hixie> | or use a different browser |
| 23:10 | <mpilgrim> | i'm using firefox 3.6 on windows |
| 23:10 | <mpilgrim> | though i assume firefox on other platforms offers an identically bad experience |
| 23:10 | <Hixie> | browsers can show whatever prompts they want if they want to make their UI less than ideal |
| 23:11 | <Hixie> | it's a tradeoff they have to make between convenience and giving the appearance of... something? not sure what the dialog helps with in this case, privacy? security? dunno. |
| 23:13 | <mpilgrim> | presumably the site could use it as a super-tracking cookie? i don't know how useful it would be since it wouldn't be accessible cross-domain |
| 23:13 | <Hixie> | it can set cookies without prompting, too |
| 23:13 | <mpilgrim> | indeed |
| 23:14 | <Hixie> | i agree that if you enable the cookie dialog, then it makes sense to prompt for appcache, web storage, web db, etc |
| 23:14 | <Hixie> | probably should also prompt with every link click, since the IP address, User-Agent string, and other parts of the HTTP message can more or less uniquely identify users even without cookies |
| 23:17 | <mpilgrim> | so, assuming i wanted to serve different static HTML files based on cookies |
| 23:17 | <mpilgrim> | i'd need a CGI script that set the cookie |
| 23:17 | <Hixie> | or javascript |
| 23:17 | <mpilgrim> | ok |
| 23:17 | <mpilgrim> | expires/caching headers set to 0 so HTML pages would always refresh from the server |
| 23:18 | <mpilgrim> | a Vary: Cookie header |
| 23:18 | <mpilgrim> | two copies of each .html file (one with the manifest attribute, one without) |
| 23:18 | <mpilgrim> | and some mod_rewrite magic that served the appropriate .html file based on the cookie in the HTTP request headers |
| 23:19 | <Hixie> | or just put the manifest on every file and make the manifest itself be the only one with magic |
| 23:19 | mpilgrim | is suddenly overwhelmed with the memory of the classic "and then a miracle occurs -->" tshirt |
| 23:19 | <mpilgrim> | no, i'm pretty sure the presence of the manifest attribute is what triggers firefox's infobar |
| 23:20 | <Hixie> | oh, right, if you want to work around firefox's ui issues you'd probably have to do that, yes |
| 23:20 | <mpilgrim> | EVERYONE WILL WANT TO WORK AROUND FIREFOX'S UI ISSUES |
| 23:21 | <Hixie> | if everyone hates firefox's ui, i imagine it'll change |
| 23:21 | <mpilgrim> | I AM JUST THE FIRST PERSON WHO HAS EVER TRIED |
| 23:21 | <Hixie> | i would recommend filing a bug with them explaining that the prompt doesn't really do anything helpful rather than yelling here :-) |
| 23:21 | <mpilgrim> | mobile gmail already does UA sniffing and only presents the offline-capable version to supported browsers (based on HTTP_USER_AGENT, i assume) |
| 23:22 | <mpilgrim> | and everyone else is still using gears |
| 23:22 | <mpilgrim> | (where "everyone else" includes zoho, rememberthemilk, and wordpress) |
| 23:23 | <mpilgrim> | (and the "supported browsers" that mobile gmail sniffs are all based on webkit and don't present any stupid UI prompts) |
| 23:23 | <mpilgrim> | (so it is quite likely that this problem has literally never come up) |
| 23:23 | <annevk> | othermaciej, I just use my inbox |
| 23:24 | <annevk> | othermaciej, so no |
| 23:26 | <mpilgrim> | i suppose my firefox bug report would be stronger if I blindly offline-enabled every page of my site and then included copies of the hate mail i expect to get from people who think my site is taking over their computer |
| 23:28 | <annevk> | this peer-to-peer stuff is interesting |
| 23:29 | <annevk> | are we taking on IM? :) |
| 23:29 | <Hixie> | annevk: dunno, i just threw IM in there because i noticed most if not all video conferencing systems also do peer-to-peer IM during the connection |
| 23:31 | <annevk> | so the way this API works is that you negotiate a unique identifier via a third-party server to establish a peer-to-peer connection? |
| 23:32 | <Hixie> | no third-party server in theory |
| 23:33 | <Hixie> | it depends on what protocol we end up building this on, but getAddress() e.g. could just return an IP address and port |
| 23:33 | <Hixie> | (of the user's machine) |
| 23:33 | <annevk> | but how do you give that to a friend? |
| 23:34 | <Hixie> | oh well i assume you're talking to them somehow already, e.g. you have a websocket connection to the server of the page the script is from |
| 23:34 | <Hixie> | and they do too |
| 23:34 | <Hixie> | e.g. for chatting in-game |
| 23:34 | <Hixie> | you would both be connected to the game server |
| 23:34 | <annevk> | right, so you have a third-party server |
| 23:34 | <Hixie> | well it's more a second-party server, but yeah |
| 23:35 | <Hixie> | (user, server, other player -- the other player is the third party) |
| 23:35 | <Hixie> | you could also just sent it by e-mail or something :-) |
| 23:35 | <annevk> | sure or you could phone the guy |
| 23:35 | <Hixie> | yeah |
| 23:36 | <Hixie> | but generally i don't think these address strings will be pretty |
| 23:36 | <annevk> | but that's not really going to happen I think :) |
| 23:36 | <annevk> | right |
| 23:36 | <Hixie> | you could imagine if we do NAT traversal that the string will include all kinds of stuff for how to poke holes in the firewalls |
| 23:36 | <Hixie> | but again, that depends on whatever we decide to build this on |
| 23:38 | <Hixie> | annevk: btw would be good to get opera's feedback on http://lists.whatwg.org/htdig.cgi/whatwg-whatwg.org/2010-February/025280.html |
| 23:38 | <annevk> | i didn't quite get it |
| 23:39 | <Hixie> | hsivonen: your feedback would likely also be useful (on http://lists.whatwg.org/htdig.cgi/whatwg-whatwg.org/2010-February/025280.html ) |
| 23:39 | <Hixie> | annevk: that's probably good feedback to send :-) |
| 23:40 | <annevk> | btw, do I misunderstand the subprotocol idea or do the people on hybi misunderstand it? |
| 23:40 | <annevk> | maybe both :) |
| 23:40 | <Hixie> | there's certainly been a lot of confusion on that issue |
| 23:40 | <Hixie> | not sure why |
| 23:41 | <Hixie> | i don't remember seeing anything from you about it so no idea if you understand it or not :-) |
| 23:43 | <franksalim> | annevk, how do you understand it? |
| 23:45 | <annevk> | I don't |
| 23:45 | <annevk> | e.g. "I would like the toString method to be called with 'html '" |
| 23:46 | <annevk> | but toString() has no arguments in this context |
| 23:46 | <annevk> | afaik |
| 23:46 | <franksalim> | where is that? |
| 23:46 | <annevk> | oh, are you talking about the subprotocol thingie? |
| 23:47 | <franksalim> | oh yes, i meant subprotocol. i see you were talking about http://lists.whatwg.org/htdig.cgi/whatwg-whatwg.org/2010-February/025280.html |
| 23:49 | <annevk> | well, I think the idea is that the subprotocol indicates how the data transmitted in the frame types is supposed to be processed, but I don't think the idea is that you can simply use all the frame types if you have a custom subprotocol |
| 23:50 | <franksalim> | I think that is correct |
| 23:51 | <franksalim> | Ports usually indicate the protocol, but because ws:// does not use host/port addressing, 'subprotocol' indicates the protocol instead |
| 23:52 | <franksalim> | Either new protocols or existing TCP protocols bound to ws |
| 23:52 | <annevk> | i.e. I am pretty certain we end up having distinct frame types for ByteArray and Stream and that a subprotocol cannot change the meaning of that |
| 23:53 | <franksalim> | annevk, so you're pretty sure we need frame types and not just a bidirectional byte stream like TCP? |
| 23:54 | <annevk> | yeah, exposing TCP to JavaScript directly won't fly |
| 23:55 | <franksalim> | for technical reasons or other reasons? |
| 23:58 | <annevk> | security, mostly |
| 23:58 | <franksalim> | I would be perfectly happy with TCP in JavaScript after the WebSocket handshake |
| 23:58 | <annevk> | oh I see |
| 23:58 | <annevk> | interesting thought |
| 23:59 | <annevk> | the reason for that is that JavaScript prolly works better being event-based |
| 23:59 | <annevk> | but I'm on shaky grounds here |
| 23:59 | <annevk> | Hixie would know |
