| 09:43 | <annevk> | mikeday, what error? |
| 09:51 | <mikeday> | annevk, UTF-32 BOM detection |
| 09:51 | <mikeday> | will always misidentify UTF-32 as UTF-16 |
| 09:51 | <mikeday> | see my post "Drop UTF-32" on whatwg mailing list |
| 09:53 | <annevk> | ah right |
| 09:55 | <mikeday> | my HTML parser can now read and skip the BOM, although it can't yet do anything else :) |
| 09:59 | <annevk> | fixed |
| 10:00 | <mikeday> | that's good. I still think UTF-32 should be verboten, though :) |
| 10:01 | jeremyb | looks up verboten |
| 10:02 | <mikeday> | it's an IETF term |
| 10:02 | <mikeday> | CAN, MAY, SHOULD, MUST, and VERBOTEN |
| 10:02 | <mikeday> | I kid. |
| 10:03 | <annevk> | :p |
| 10:04 | <mikeday> | but seriously, who in their right mind would use it for bits over the wire? |
| 10:04 | <mikeday> | I mean, you might use it as an in-memory representation, perhaps. |
| 10:04 | <mikeday> | but sending it over TCP/IP is just nuts |
| 10:04 | <annevk> | in-memory people use UTF-16 |
| 10:05 | <mikeday> | well, we use UTF-8 |
| 10:05 | <mikeday> | but some people do use UTF-32, if they use C with wchar == int |
| 10:05 | <mikeday> | so it's not entirely unheard of. |
| 10:06 | <mikeday> | but no one uses it for web pages, nor should they, so BAN IT :) |
| 10:06 | <annevk> | isn't UTF-16 way easier to traverse over the characters without having to check them each? |
| 10:06 | <mikeday> | you still have surrogates in UTF-16 |
| 10:07 | <mikeday> | it depends what you're doing with the characters |
| 10:07 | <mikeday> | UTF-8 has some nice properties |
| 10:07 | <mikeday> | and if you're doing binary comparison of strings, you can just check bytes, not characters, and it's faster |
| 10:07 | <mikeday> | the choice of encoding to use for in-memory representations is quite specific to your use cases and environment. |
| 10:08 | <mikeday> | Linux tends to use UTF-8, Windows tends to use UTF-16 |
| 10:08 | <jeremyb> | mac? |
| 10:08 | <jeremyb> | solaris? posix? |
| 10:09 | mikeday | shrugs |
| 10:09 | <mikeday> | the unixy bits of MacOS X probably UTF-8, I don't know about higher layers though |
| 10:11 | <mikeday> | I'm not sure why Windows went with UTF-16, but could be related to making Microsoft Word work well in Japan |
| 10:11 | <jeremyb> | just japan or all of CJK? |
| 10:11 | <mikeday> | all of it, but no one was paying for software in China at the time :) |
| 10:12 | <mikeday> | UTF-8 isn't optimal for CJK, and using it would have been a disadvantage compared to local apps using JIS or whatever |
| 10:12 | <mikeday> | and coding up two versions of Word for each version was becoming a nuisance... |
| 10:12 | mikeday | shrugs |
| 10:12 | <mikeday> | just a guess. |
| 10:13 | <mikeday> | For UNIX systems the weight of experience with C string processing is enough to make UTF-8 sound more attractive |
| 10:13 | <mikeday> | and who knows, maybe increased processor speed and memory make the down side less of a problem for non-Latin users |
| 10:14 | <gsnedders> | Mac OS X uses UTF-16 for Cocoa, and NFKC UTF-8 for the POSIX layer |
| 10:14 | <mikeday> | right, that's what I figured |
| 10:14 | <gsnedders> | actually, that's wrong. |
| 10:14 | <gsnedders> | it isn't NFKC… |
| 10:14 | <mikeday> | right, then I'm wrong :) |
| 10:15 | <gsnedders> | UTF-8-MAC |
| 10:15 | <mikeday> | hmm, that's not another broken UTF-8 variant is it? |
| 10:15 | <gsnedders> | http://developer.apple.com/qa/qa2001/qa1173.html |
| 10:15 | <gsnedders> | mikeday: some characters are forbidden |
| 10:16 | <gsnedders> | mikeday: and NFD |
| 10:16 | <mikeday> | so it's just a pure subset of UTF-8? |
| 10:16 | <gsnedders> | mikeday: it's really just NFD UTF-8, but it MUST be NFD. no other normalisation form is allowed. |
| 10:16 | <mikeday> | that's not so bad then |
| 10:17 | <mikeday> | not really an "encoding" in the XML/HTML sense, I suppose |
| 10:17 | <mikeday> | even if it could be used over the wire, eg. with SAMBA or NFS |
| 10:19 | <gsnedders> | UTF-8 NFD is used for FS, UTF-16 is used for Core Foundation, and I don't know what else is allowed where |
| 10:20 | <mikeday> | we use UTF-8 during font enumeration via ATS |
| 10:21 | <gsnedders> | Carbon, I think, uses UTF-8 |
| 10:22 | <mikeday> | do you have an opinion on the question of forbidding UTF-32 from HTML? :) |
| 10:22 | <gsnedders> | mikeday: see my email |
| 10:22 | mikeday | checks |
| 10:22 | <gsnedders> | <http://lists.whatwg.org/pipermail/whatwg-whatwg.org/2007-May/011330.html>; |
| 10:22 | <mikeday> | true, the implementors burden is not huge |
| 10:23 | <mikeday> | but I gave other reasons as well :) |
| 10:23 | <gsnedders> | Forbidding it serves no practice use. |
| 10:23 | <mikeday> | neither does allowing it |
| 10:24 | <mikeday> | which is in effect compelling user agents to support it, or risk appearing non-compliant, |
| 10:24 | <mikeday> | even if the spec technically doesn't require them to support it |
| 10:24 | <gsnedders> | If we start off by accepting any character set is allowed, what advantages does banning it give us, and what disadvantages do we have? |
| 10:24 | <mikeday> | it also makes it look as if using UTF-32 for web pages is not an insane thing to do :) |
| 10:25 | <annevk> | mikeday, last I heard implementing it is not that difficult... |
| 10:25 | <gsnedders> | but they're no more compelled to support it than any other of the many encodings registered |
| 10:25 | <gsnedders> | annevk: it's _VERY_ easy to implement |
| 10:26 | <mikeday> | annevk, how about implementing it *correctly* :P |
| 10:26 | <gsnedders> | mikeday: still _VERY_ easy |
| 10:27 | <annevk> | mikeday, you implement a very slick C parser for HTML and I'll add the UTF-32 support :p |
| 10:27 | <annevk> | (with very slick I also mean fast and such :) ) |
| 10:27 | <mikeday> | the HTML5 spec also provides two BOMs for UTF-32, while the XML spec suggests 4, including unusual byte orders |
| 10:28 | <annevk> | Prolly copied from some implementation... |
| 10:29 | <jeremyb> | what's a BOM? |
| 10:29 | <annevk> | byte order mark |
| 10:29 | <mikeday> | better to say that if UTF-32 is used, it should be specified by name, not automagically identified by BOM |
| 10:29 | <jeremyb> | ahh |
| 10:29 | <gsnedders> | most implementations will have some method to create characters in the internal representation of the characters from a unicode code-point, and then you can just read it from a BE/LE stream |
| 10:29 | <gsnedders> | mikeday: UTF-32 is unclear in name, can be BE or LE |
| 10:29 | <mikeday> | as FE FF 00 00 is almost certainly UTF-16 with a null character, not UTF-32 |
| 10:30 | <gsnedders> | U+00 is disallowed in HTML |
| 10:30 | <gsnedders> | converted to the code replacement point in the input stream |
| 10:30 | <annevk> | jeremyb, it is an optional set of bytes at the start of a byte stream that identifies how to interpret the rest of the stream |
| 10:30 | <mikeday> | yes, but in practice I bet it's more common to find pages with nulls than pages using UTF-32 |
| 10:31 | <gsnedders> | mikeday: I'd bet otherwise :P |
| 10:31 | <mikeday> | especially pages generated by CGI or whatever |
| 10:31 | <mikeday> | really?? have you ever seen a page written in UTF-32? Or are you writing one right now? :) |
| 10:31 | <jeremyb> | annevk: right... this is processor dependant, right? |
| 10:32 | <gsnedders> | mikeday: I've seen pages with U+00 and UTF-32 pages. I think the latter would be more common |
| 10:32 | <annevk> | jeremyb, I'm not sure I understand that question |
| 10:32 | <gsnedders> | jeremyb: the order of the bytes of the BOM shows whether it is big/little/middle endian |
| 10:32 | <gsnedders> | jeremyb: (as I assume you are referring to endianness) |
| 10:32 | <mikeday> | gsnedders, oh, I've only seen the former, never the latter. Have to ask Hixie to check the google cache for a definitive resolution on this one, I guess :) |
| 10:32 | <gsnedders> | mikeday: most probably. |
| 10:32 | <jeremyb> | never heard of middle before. yes i was talking about endianness |
| 10:33 | <gsnedders> | mikeday: more to the point, how many UTF-16 docs start with BOM then U+00 |
| 10:33 | <mikeday> | true, that must narrow it down somewhat :) |
| 10:33 | <gsnedders> | jeremyb: middle is rare nowadays, but exists. more common in other subject matters (like middle endian dates, which start with the month) |
| 10:33 | <mikeday> | as not many people use UTF-16 on the web either :) |
| 10:33 | jeremyb | supposes that he should read up on endianness |
| 10:34 | <gsnedders> | sometimes I feels odd telling people such things knowing that almost certainly everyone here is older than me :P |
| 10:34 | jeremyb | avoids middle endian dates |
| 10:34 | <mikeday> | dates could do with a BOM as well |
| 10:34 | <jeremyb> | how old are you? |
| 10:34 | gsnedders | had to use little endian dates in his exam |
| 10:34 | <gsnedders> | jeremyb: 15 |
| 10:34 | <mikeday> | sometimes it's hard to tell, eg. with 02/03/04 |
| 10:34 | <gsnedders> | *exams |
| 10:35 | <gsnedders> | mikeday: HAH! D/M/Y as the first char? |
| 10:36 | <jeremyb> | gsnedders: i know how you feel although i'm not quite that young |
| 10:36 | <gsnedders> | jeremyb: I've been around here since I was… 13 maybe? |
| 10:37 | <gsnedders> | but near my 14th birthday |
| 10:38 | <jeremyb> | you've been on IRC longer than i have! |
| 10:39 | <gsnedders> | Oh, I've in IRC longer. |
| 10:39 | <gsnedders> | just nowhere near WHATWG |
| 10:39 | <gsnedders> | though mainly all I talk about on IRC is my lust life :P |
| 11:48 | <mikeday> | I noticed that Sam Ruby has made a HTML parser in... Ruby |
| 11:48 | <annevk> | at some point it'll support <ruby> too |
| 11:49 | <mikeday> | hah! |
| 11:49 | <mikeday> | that would be so awesome :) |
| 11:50 | <mikeday> | I wonder if anyone has ever tried using <ruby> equivalent to <script language="ruby"> |
| 11:51 | <othermaciej> | we need Ruby's ruby <ruby> parser |
| 11:51 | <mikeday> | hmm, sung to the tune of "Ruby ruby" |
| 11:51 | <mikeday> | http://www.oldielyrics.com/lyrics/dion_and_the_belmonts/ruby_baby.html |
| 11:53 | krijnh | adds it to http://krijnhoetmer.nl/lyrics/ :+ |
| 11:55 | <Jero> | http://lyrics.doheth.co.uk/songs/kaiser-chiefs/yours-truly-angry-mob/ruby.php > that one |
| 11:57 | <mikeday> | guess perl and python don't have that many songs about them |
| 11:59 | <Jero> | does Perl Jam count? >_> |
| 11:59 | <Jero> | and don't forget that Perl has its own Pokemon game |
| 12:03 | <Jero> | oh wait, that should be Pearl of course |
| 12:04 | <annevk> | Python has Monty Python |
| 12:04 | <annevk> | that should be enough |
| 12:06 | <Jero> | annevk: how's the university of utrecht? |
| 12:07 | <annevk> | I'm not there much, but it's pretty cool |
| 12:07 | <Jero> | you're doing computer science, right? |
| 12:08 | <annevk> | I'm actually doing information science |
| 12:08 | <annevk> | but I'm planning to follow all courses of computer science that seem interesting (and am already doing so) |
| 12:09 | <Jero> | oh wow :p |
| 12:10 | <annevk> | (I'm "skipping" this semester though. Doing a research project instead.) |
| 12:10 | annevk | is in Oslo because of that |
| 12:10 | <Jero> | sweet |
| 12:11 | <Jero> | i'm still sort of deciding what to do next year |
| 12:12 | Jero | also lives in the Netherlands |
| 12:12 | gsnedders | is listening to Ruby by Kaiser Chiefs from Yours Truly, Angry Mob |
| 12:12 | <gsnedders> | that wouldn't be so scary if you weren't talking about that a few minutes ago. |
| 12:26 | Philip` | wonders if it'd be interesting to make an HTML compressor, that uses all the error handling rules of HTML5 to make totally non-conforming documents that get parsed exactly the same as the original and are as small as possible |
| 12:29 | <peepo> | svgz? |
| 12:29 | <peepo> | htmlz... |
| 12:29 | <annevk> | The idea of having error handling defined is that we can modify bits of it in due course... |
| 12:32 | <annevk> | Although not lightly, of course... |
| 12:41 | <Jero> | If there is no furthest block, ... pop all the nodes ... from the current node up to the formatting element... |
| 12:41 | <Jero> | just to be sure, that doesn't include the formatting element, right? |
| 12:43 | <annevk> | don't think so |
| 12:44 | <annevk> | heh |
| 12:44 | <annevk> | exactly 42 web forms 2 repetition tests |
| 12:44 | <annevk> | the forms dir on tc.labs.opera.com/html/ contains 224 tests |
| 12:47 | <annevk> | So I just read http://www.456bereastreet.com/archive/200705/browsers_will_treat_all_versions_of_html_as_html_5/ again. Why would changing our behavior for a new version of HTML make stuff easier for developers? |
| 12:47 | <annevk> | We'll have a new set of bugs specific to the new version and transitioning old content to HTML5 becomes harder, etc. |
| 12:48 | <annevk> | There are so many obvious problems for developers and yet they don't seem to realize them?? |
| 13:28 | <karlUshi> | python joke. cf the previous discussions |
| 13:28 | <karlUshi> | http://www.redmountainsw.com/wordpress/archives/a-joke-in-the-python-interpreter |
| 13:45 | <annevk> | Philip`, is it easy to modify your survey tool to check how often the adoption agency algorithm is triggered? Especially the clone node operation in it... |
| 13:51 | <Philip`> | If there's somewhere in the html5lib code where it could poke some value that I could read out again after parsing, then I guess that'd be easy enough |
| 13:54 | <Philip`> | (But "survey tool" makes it sound more impressive and useful than just being the non-representative collection of a couple of thousand pages with a parser being called on them in a loop which is all it really is :-) ) |
| 14:00 | <annevk> | hehe |
| 14:00 | <annevk> | I think you'd have to make such a value yourself |
| 14:06 | <Jero> | 6. Let a bookmark note the position of the formatting element in the list of active formatting elements... |
| 14:06 | <Jero> | hmm, I guess that line should be rephrased a little |
| 14:06 | <Jero> | (step 6 of the adoption agency) |
| 14:08 | <Jero> | oh nvm, i understand it now, note is used as a verb here |
| 14:19 | <Philip`> | annevk: I ought to have time this afternoon to try looking at that |
| 14:21 | <annevk> | cool |
| 14:21 | <annevk> | Jero, btw, I'd aim for making a full parser at first and create filtering tools for it later |
| 14:22 | <annevk> | Jero, that makes it more usable for other people who want to use it where support for <blah> and <link> is important |
| 14:22 | <Jero> | yeah, that's sort of what i'm doing now |
| 14:22 | <annevk> | cool :) |
| 14:22 | <Jero> | but first i need the adoption agency working properly |
| 14:25 | <Philip`> | It seems worrying that the adoption agency sometimes cuts its children into multiple pieces |
| 14:26 | <annevk> | it clones them |
| 14:26 | <annevk> | and distributes them among parents afterwards |
| 14:27 | <Philip`> | But the content of the original tags gets split apart and spread between all the clones |
| 14:28 | <Philip`> | I'm sure that kind of thing can't be legal |
| 14:29 | <annevk> | :p |
| 14:37 | <gsnedders> | I need to do something about the tree construction… but probably better to get the other stuff working fix :P |
| 14:37 | <gsnedders> | *first |
| 14:41 | <Jero> | where in the algorithm do P elements close elements such as A and B? |
| 14:41 | <annevk> | why would <p> close them? |
| 14:42 | <annevk> | <a><p>test</p></a> is legal |
| 14:42 | <Jero> | or really, sweet |
| 14:42 | <annevk> | well, "legal" |
| 14:42 | <annevk> | <b><p>test</p></b> should work too |
| 14:42 | <Jero> | yeah, i know what you mean :p |
| 14:43 | <annevk> | <b><p>test</b></p> makes things interesting... |
| 14:45 | <Jero> | hmm, but when I look at the testcase with the code "<a><p>X<a>Y</a>Z</p></a>" in http://html5lib.googlecode.com/svn/trunk/tests/tree-construction/tests1.dat , the parser closes the first A element when it sees the P element |
| 14:45 | <annevk> | you make incorrect assumptions |
| 14:46 | <annevk> | the magic happens the moment you hit the second <a> |
| 14:46 | <annevk> | not when you hit <p> |
| 14:46 | <Jero> | ah |
| 15:08 | annevk | curses the internal subset |
| 15:15 | <annevk> | FWIW: We won't drop support for UTF-32 |
| 15:19 | <zcorpan_> | annevk: why not? |
| 15:19 | <annevk> | There's no good reason not to support it |
| 16:24 | <annevk> | http://www.andybudd.com/archives/2007/05/xtech_2007/ |
| 16:31 | <Dashiva> | "other interface languages like MXML and ZUL" |
| 16:32 | <annevk> | yeah... |
| 16:37 | <Philip`> | annevk: Which clone node operation do you mean? On 2274 pages, I see the clone in step 7.5 being called at least once on 5 pages, and the one in step 9 on 149 pages |
| 16:48 | <annevk> | can you count how many times they are being hit? |
| 16:51 | <Philip`> | The 7.5 one was 7 times, the 9 one was 691 times |
| 16:53 | <Philip`> | (Worst offender: http://www.ed.gov/parents/needs/speced/iepguide/index.html which runs that step 62 times) |
| 16:56 | <Philip`> | (I assume it's the stuff like "<i><b><p>Parents</b></i> are key members of the IEP team.") |
| 17:04 | <annevk> | ouch |
| 17:05 | <annevk> | that's a lot of additional node |
| 17:05 | <annevk> | s |
| 17:06 | <annevk> | although it's less than one node per page extra on average |
| 17:06 | <annevk> | would be nice if Hixie could test this |
| 17:19 | <zcorpan_> | not much has been posted at the whatwg blog lately |
| 17:23 | <Dashiva> | Except those spam posts :) |
| 19:12 | annevk | goes to order a t-shirt |
| 19:16 | <zcorpan_> | annevk: a black one or a white one? :) |
| 19:18 | <annevk> | black |
| 19:18 | <annevk> | 25 dollars |
| 19:19 | <annevk> | better be good |
| 19:19 | <annevk> | and it better arrives next week... |
| 19:19 | <zcorpan_> | a 5 > 2 t-shirt? |
| 19:19 | <annevk> | (I guess that's unlikely) |
| 19:19 | <annevk> | yeah |
| 19:20 | <annevk> | including shipping cost |
| 19:20 | <zcorpan_> | i got mine after 3 days iirc |
| 19:20 | <annevk> | sounds good |
| 19:20 | <annevk> | that means I can actually wear it when I'm speaking |
| 19:20 | <annevk> | XML5: 47 DOCTYPE tokenizer states and 71 in total |
| 19:21 | <Dashiva> | How about just... pretending there's no such thing |
| 19:21 | <Dashiva> | Go bogus comment immediately |
| 19:21 | <annevk> | Ubuntu: doesn't install on a T60 Widescreen |
| 19:21 | <Dashiva> | Would that destroy roundtripping, maybe |
| 19:22 | <annevk> | Would destroy default attribute values |
| 19:22 | <annevk> | Would also destroy entities... |
| 19:22 | <Dashiva> | No, once you see <!doctype I mean |
| 19:23 | <Dashiva> | Oh wait, nm. Silly |
| 19:26 | <annevk> | zcorpan_, when is your presentation? |
| 19:27 | <zcorpan_> | wednesday |
| 19:27 | <annevk> | you'll put it online at some point? |
| 19:27 | <zcorpan_> | yeah |
| 19:27 | <zcorpan_> | http://simon.html5.org/temp/html5-geekmeet.html |
| 19:28 | <annevk> | oh, cool |
| 19:28 | annevk | looks |
| 19:28 | <zcorpan_> | some things are commented out |
| 19:28 | <annevk> | lang=sv |
| 19:28 | <annevk> | ah |
| 19:28 | <zcorpan_> | i'll expand on canvas and wf2 too |
| 19:28 | <zcorpan_> | though i don't know much about canvas :| |
| 19:29 | <annevk> | that looks quite neat |
| 19:29 | <zcorpan_> | thanks |
| 19:30 | <zcorpan_> | i might translate the slides to english afterwards |
| 19:30 | <annevk> | my next presentation will be "inspired" by yours ;) |
| 19:30 | <Lfe> | zcorpan_: where is geekmet this wednesday? stockholm? |
| 19:30 | <zcorpan_> | Lfe: yeah |
| 19:30 | <zcorpan_> | http://www.robertnyman.com/2007/04/26/geek-meet-may-2007-html-5-and-xhtml/ |
| 19:31 | <Lfe> | darn, missed that one :( |
| 19:31 | <annevk> | zcorpan_, so Opera actually parses this HTML5 crap correctly? :) |
| 19:31 | <zcorpan_> | annevk: yep :) |
| 19:31 | annevk | used XHTML5 so far to avoid the issue |
| 19:31 | <zcorpan_> | Lfe: it's not too late to sign up |
| 19:33 | <zcorpan_> | annevk: omitting optional tags in combination with the new tags doesn't quite work |
| 19:33 | <annevk> | yeah ok |
| 19:33 | <zcorpan_> | but otherwise it works fine |
| 19:33 | <zcorpan_> | <datalist><option></datalist> for instance (think i've filed a bug about that one) |
| 19:34 | <annevk> | yeah |
| 19:34 | <annevk> | note that the parsing algorithm doesn't define stuff like that so far... |
| 19:36 | <zcorpan_> | right |
| 19:36 | <annevk> | (not sure who I'm talking to here, everybody knows that :) ) |
| 19:38 | <zcorpan_> | i'm supposed to explain how <canvas> works. i need some help with that :) |
| 19:38 | <Lfe> | zcorpan_: true, but i don't live there.. so kinda hairy to make all the arrangements :( |
| 19:39 | <zcorpan_> | Lfe: where do you live? |
| 19:39 | <Lfe> | zcorpan_: kalmar |
| 19:39 | <zcorpan_> | ok |
| 19:40 | <annevk> | zcorpan_, you select the element, get the 2d context on it and draw stuff |
| 19:41 | zcorpan_ | makes notes |
| 19:41 | <annevk> | search for <canvas in http://lists.w3.org/Archives/Public/www-archive/2007May/att-0063/html5-short.xml |
| 19:42 | <zcorpan_> | annevk: thanks |
| 19:42 | <annevk> | http://lists.w3.org/Archives/Public/www-archive/2007Apr/att-0036/html5.xml has a simple drawing application using <canvas> |
| 19:43 | <annevk> | (although you can't actually see it in non-presentation mode) |
| 19:43 | <zcorpan_> | yup. i was thinking of showing some cool canvas demos at the presentation, though i won't embed them in the actual presentation |
| 19:43 | <zcorpan_> | probably just link to them and switch between tabs |
| 19:44 | <annevk> | for my next presentation I'd like to show the doom game |
| 19:44 | <zcorpan_> | yeah |
| 19:45 | <annevk> | btw, when will you start? |
| 19:45 | <annevk> | not until July or earlier? |
| 19:45 | <zcorpan_> | at opera? |
| 19:45 | <zcorpan_> | week 25 |
| 19:45 | <annevk> | when is that? :) |
| 19:46 | <zcorpan_> | http://vecka.nu/ |
| 19:46 | <annevk> | ok, so around the 20th of June |
| 19:46 | <zcorpan_> | 18 june |
| 19:46 | <zcorpan_> | yeah |
| 19:46 | <annevk> | cool |
| 19:46 | <Lfe> | i made a widget for that some months ago, it's insane how hard it is to find the current week num :) |
| 19:47 | <annevk> | maybe I should try to go to Sweden for a week or so |
| 19:47 | <zcorpan_> | i'll probably quit studying after the summer and start to work instead |
| 19:47 | <zcorpan_> | annevk: when? |
| 19:47 | <annevk> | when you're around, of course :) |
| 19:48 | <annevk> | i meant to the Linkoping office |
| 19:48 | <zcorpan_> | ok |
| 19:48 | <annevk> | or will you go to Oslo? |
| 19:48 | <zcorpan_> | no |
| 19:48 | <annevk> | right |
| 19:49 | <zcorpan_> | i'll be in linköping the first week i think, then i'll only be there occasionally |
| 19:49 | <annevk> | ah ok |
| 19:50 | <hasather> | annevk: will you be in Oslo most of the summer? |
| 19:50 | <zcorpan_> | would be cool to meet, then i can practice my dutch a bit too :) |
| 19:51 | <annevk> | hasather, yeah, sort of |
| 19:51 | <annevk> | hasather, as in, when I'm not travelling I'll be in Oslo |
| 19:51 | annevk | is in Oslo right now, fwiw |
| 19:52 | <hasather> | ok, I'll start June 11th |
| 20:08 | <hasather> | annevk: what kind of problems did you have with Ubuntu btw? Works nicely on my Z60m, but I don't know how much differs between them |
| 20:16 | <Philip`> | Someone just needs to implement an HTML5 parser in JavaScript, and then you could do something like in http://canvex.lazyilluminati.com/misc/sexp.html to enable the new parsing algorithm in legacy browsers |
| 20:43 | zcorpan_ | ponders whether the html5 spec should distinguish between "element" and "element type" |
| 20:47 | <hasather> | zcorpan_: preferrably, but 'element' is probably clear enough. That's what most people use to refer to both terms |
| 20:48 | <zcorpan_> | yeah. i guess it isn't really ambiguous in context |
| 23:48 | <zcorpan_> | hm, how do we solve the gmail problem of not being able to use real links because of Referer? use a new attribute <a secrethref="">? |
| 23:49 | <othermaciej> | zcorpan_: maybe just a noreferrer attribute? |
| 23:49 | <othermaciej> | I would guess only <a> needs it |
| 23:49 | <Hixie> | and <img> |
| 23:49 | <othermaciej> | for <script> or <img> it could be a security issue to skip referrer |
| 23:50 | <zcorpan_> | othermaciej: but then legacy UAs would still send the Referer |
| 23:50 | <othermaciej> | for <img> you clearly don't want it, since it would make hotlinking undetectable |
| 23:50 | <Hixie> | <img> in an HTML e-mail shouldn't send Referer headers |
| 23:50 | <Hixie> | (welcome to conflicting requirements) |
| 23:50 | <othermaciej> | <img> to an external resource in HTML e-mail shouldn't display by default, and possibly not at all |
| 23:51 | <othermaciej> | just pinging the server back at all is a privacy concern |
| 23:51 | <Hixie> | not by default, sure |
| 23:51 | <othermaciej> | Referer doesn't really add much privacy violation beyond just loading the <img> |
| 23:51 | <Hixie> | depends what's in the url |
| 23:53 | <othermaciej> | put it this way, someone who isn't out to violate your privacy would likely embed the image rather than making it an external reference, since in most mail clients these days you have to go out of your way to do the latter; and if they are out to violate your privacy, they could do much worse by careful URL design than with Referer |
| 23:54 | <Hixie> | i don't deny any of that |
| 23:54 | <Hixie> | i'm just saying there are also requests for not passing referrers for <img>s |
| 23:54 | <Hixie> | whether we satisfy those requests or not is another question |
| 23:54 | <zcorpan_> | <a>link text <href>http://...</href></a> -- that would be backwards compatible |
| 23:55 | <Hixie> | kinda |
| 23:55 | <zcorpan_> | you could script that for legacy UAs to fake the link |
| 23:56 | <zcorpan_> | or just leave it and let the user copy the URL |
| 23:56 | <Hixie> | you could script the other one too |
| 23:56 | <zcorpan_> | yeah, but the user wouldn't know the URL without script |
| 23:58 | <zcorpan_> | how does gmail handle links in its "html basic" version? |
| 23:58 | <Hixie> | no i mean <a href="..." noreferer> could be scripted |
| 23:58 | <Hixie> | dunno, try it |
| 23:58 | <zcorpan_> | yeah but then legacy UA would still send Referer... isn't that an issue? |