| 08:02 | <duryodhan> | for the color chooser, maybe one could define the color ends like#ff0 and #fff , and all shades in between will be shown for the user to pick, instead of writing a list of all colors |
| 08:02 | <duryodhan> | not sure how usefull that will be though... :) |
| 13:39 | <Philip`> | I get a parse error for "unrecognised entity name" on 60% of pages, which seems quite surprising |
| 14:08 | <Philip`> | Oh, right, the unrecognised entities were almost all in attributes, which sounds like <a href="a?b&c">, so that's not so surprising |
| 14:10 | <Philip`> | 60% of pages have that error, 30% have no parse errors at all, 10% have duplicate attributes (odd?), 7% have non-permitted slashes, etc |
| 14:24 | <zcorpan_> | Philip`: what are the duplicate attributes? |
| 14:26 | <zcorpan_> | things like <meta name=keywords content=the best of the best>? |
| 14:47 | <aas> | anybody here? |
| 14:49 | <zcorpan_> | yep |
| 14:53 | <aas> | i have a little proble with XHTML MP DOCTYPE declaration |
| 14:53 | <zcorpan_> | ok |
| 14:56 | <aas> | while i'm creating XHTML MP document in servlet through XSLT transformation it's connection to w3c.org and verifing content, and then throw FileNotFound exception on http://www.wapforum.org/DTD/xhtml-mobile10-model-1.mod file |
| 14:56 | <aas> | hmmm. i hope my english seems not very bad for you |
| 14:57 | <aas> | so, what should i do for preventing this error? |
| 14:57 | <zcorpan_> | don't use a doctype |
| 14:57 | <aas> | ) |
| 14:58 | <aas> | of course, it will solve my problem, but it's incorrect a little a think |
| 14:58 | <aas> | *i |
| 14:58 | <zcorpan_> | why is it incorrect? |
| 14:59 | <aas> | i read it's necessary to include doctype in document |
| 14:59 | <zcorpan_> | according to whom? |
| 14:59 | <aas> | wait a sec |
| 15:01 | <aas> | hmmm... |
| 15:03 | <aas> | then another question: a you sure that all phones can process wap2.0 page without doctype element? |
| 15:04 | <zcorpan_> | pretty sure |
| 15:04 | <zcorpan_> | http://simon.html5.org/articles/mobile-results |
| 15:05 | <aas> | are you writing wap sites? |
| 15:05 | <zcorpan_> | no |
| 15:06 | <zcorpan_> | according to my testing, most phones treat your (carefully crafted) page (regardless of what it is) pretty much like desktop browsers treat html |
| 15:06 | <aas> | he, very interesting article, thanks ) |
| 15:07 | <zcorpan_> | even if it validates as something fancy and is served as application/xhtml+xml |
| 15:07 | <zcorpan_> | it gets tag soup treatment by most phones |
| 15:07 | <zcorpan_> | (sadly) |
| 15:08 | <aas> | hmm. did you use emulators? |
| 15:09 | <zcorpan_> | i think a few emulators were tested too |
| 15:09 | <zcorpan_> | the s60 webkit emulator iirc |
| 15:09 | <zcorpan_> | (the result of which was confirmed on a real device) |
| 15:12 | <aas> | how do you think, does mobile phone browsers look at doctype at all? |
| 15:12 | <zcorpan_> | not sure. probably not |
| 15:13 | <zcorpan_> | opera might look at it |
| 15:14 | <zcorpan_> | if you serve the same markup to desktop browsers, and want to avoid quirks mode, then you may want to use some doctype |
| 15:14 | <zcorpan_> | (if you serve it as text/html) |
| 15:15 | <aas> | problem is that i already wrote code for converting web pages... useless now |
| 15:16 | <Philip`> | zcorpan_: http://www.sh3bwah.com/ does <img border="0" ... border=0>, http://www.sapo.pt/ does <a class="subcat" ... class="url">, http://www.nate.com/ does <input name="news_radio" ... name="NS" ...>, http://www.sanook.com/ does <a ... target="_blank" ... target="_blank" ...> |
| 15:17 | <zcorpan_> | Philip`: any dups for style? |
| 15:22 | <aas> | zcorpan_ thanks a lot! it was very useful for me |
| 15:22 | <Philip`> | zcorpan_: http://www.uume.com/ |
| 15:22 | <aas> | sorry for my english |
| 15:23 | <Philip`> | <li ... style="display:none" style="background:none" ...> |
| 15:24 | <Philip`> | (I don't automatically collect data about which attributes were duplicated, so I'm just checking the list of pages through the W3C Validator to get more details) |
| 15:24 | <zcorpan_> | Philip`: ok |
| 15:24 | <zcorpan_> | Philip`: dups for style is nasty in ie |
| 15:25 | <zcorpan_> | Philip`: if we were to support what ie does with dups for style in html5, it would require hacks in the tokenizer :( |
| 15:28 | <Philip`> | <b style="color:red; /*" style="font-size:2em; */"> |
| 15:28 | <Philip`> | Looks like you can't just concatenate the attribute values :-( |
| 15:28 | <Philip`> | I'll try adding some instrumentation to see how common each specific duplicated attribute is |
| 15:29 | <zcorpan_> | indeed, you need to parse each attribute separately and then combine them |
| 15:30 | <Philip`> | <b style="color:red" style="color:green"> |
| 15:30 | <zcorpan_> | pretty much as if it was <style> b { color:red; /*</style> <style> b { font-size:2em; */</style> |
| 15:31 | <Philip`> | Have to reverse the order, it seems |
| 15:31 | <zcorpan_> | ah yep, that too |
| 15:31 | <zcorpan_> | as if it was <style> b { font-size:2em; */</style> <style> b { color:red; /*</style> |
| 15:32 | <Philip`> | <b style="color:red" style="color:green !important"> |
| 15:45 | <zcorpan_> | it is unclear to me what should happen here: http://software.hixie.ch/utilities/js/live-dom-viewer/?%3C%21DOCTYPE%20html%3E%0D%0A%3Cinput%3E%3Cscript%3Edocument.body.firstChild.tabIndex%20%3D%20%22x%22%3B%20w%28document.body.firstChild.getAttribute%28%22tabindex%22%29%29%3C/script%3E |
| 15:46 | <zcorpan_> | "On setting, the given value must be converted to a string representing the number as a valid integer in base ten and then that string must be used as the new content attribute value." |
| 16:16 | <zcorpan_> | ah. it's defined in http://dev.w3.org/cvsweb/~checkout~/2006/webapi/Binding4DOM/Overview.html?content-type=text/html;%20charset=utf-8#es-long |
| 16:17 | <zcorpan_> | very much needed spec that one |
| 16:17 | zcorpan_ | thinks |
| 16:44 | <zcorpan_> | wonder why <col> isn't allowed as a child of <table> |
| 16:44 | <Philip`> | zcorpan_: I see duplicate @style on 10 pages (out of ~480) |
| 16:44 | <Philip`> | ( http://www.uume.com http://www.megaupload.com http://www.goo.ne.jp http://www.whenu.com http://www.alibaba.com http://www.naver.com http://www.sohu.com http://www.51.com http://www.domaintools.com http://www.6rb.com ) |
| 16:45 | <zcorpan_> | Philip`: wow, that's a lot |
| 16:49 | <zcorpan_> | the first two don't seem to break anything by not doing what ie does |
| 16:49 | zcorpan_ | checks the others |
| 16:50 | <Philip`> | style="" style="" |
| 16:50 | <Philip`> | style="width:300px;" style="ime-mode: active;" |
| 16:50 | <Philip`> | style="padding:0px 0px;" style="font-size:10px;" |
| 16:50 | <Philip`> | style="float:left;width:40%;height:30px;" style="padding:0 0 4px 10px;" |
| 16:50 | <Philip`> | style="WIDTH: 317px" name=query tabindex=2 onfocus="return setTextBox(event, 0);" style="BACKGROUND-POSITION: left 50%; BACKGROUND-IMAGE: url(http://wstatic.naver.com/www/images3/cursor.gif); BACKGROUND-REPEAT: no-repeat" |
| 16:50 | <Philip`> | style=" font-weight:normal; text-decoration:underline;color:#0000cc;font-size:12px;margin-left:240px;" style="behavior: url(#default#homepage)" |
| 16:50 | <Philip`> | style="FONT-SIZE: 12px" style="color:#ff0000" |
| 16:50 | <Philip`> | style="line-height: 1.0em;" style="vertical-align: top" |
| 16:50 | <Philip`> | style="font-family: Tahoma; font-size: 8pt; border: 1 solid #808080" style="font-family: Tahoma; font-size: 8pt" |
| 16:51 | <Philip`> | (Those are the pairs I see on one element) |
| 16:51 | <Philip`> | (for the ten sites) |
| 16:51 | <zcorpan_> | Philip`: thanks |
| 16:54 | <Philip`> | Three of those pages have style="behaviour:url(#default#homepage)", which seems unusual |
| 16:54 | <Philip`> | and the links with those styles also have onClick="this.style.behavior='url(#default#homepage)';this.setHomePage('http://www.naver.com');"; |
| 16:54 | <Philip`> | (The one with duplicate style attributes, of which one was that #default#homepage thing, also had the onclick on the same element) |
| 16:55 | <zcorpan_> | that's an ieism we probably don't need to copy |
| 18:03 | <jgraham> | hsivonen: I just found a comment in the html5lib code for the processCharacter method in the InBody phase: |
| 18:03 | <jgraham> | # XXX The specification says to do this for every character at the |
| 18:03 | <jgraham> | # moment, but apparently that doesn't match the real world so we don't |
| 18:03 | <jgraham> | # do it for space characters. |
| 18:03 | <jgraham> | self.tree.reconstructActiveFormattingElements() |
| 18:03 | <jgraham> | So it looks like we don't reconstruct afe's for space elements by design |
| 19:03 | <hsivonen> | jgraham: hmm. testing suggests that real world requires what the spec says--not what html5lib does (Re: afe) |
| 19:04 | <jgraham> | Interesting. |
| 19:05 | <jgraham> | At least we should raise this with Hixie |
| 19:06 | <jgraham> | and ask Anne if he knows why html5lib is different here |
| 19:06 | jgraham | goes to re-enable current-spec compliant behaviour |
| 19:08 | <hsivonen> | http://software.hixie.ch/utilities/js/live-dom-viewer/?%3C%21DOCTYPE%20html%3E%3Cp%3E%3Cb%3E%3Ci%3E%3Cu%3E%3C/p%3E%20%3Cp%3EX |
| 19:09 | <hsivonen> | Firefox 2.0, WebKit trunk, Safari 2.0.4, Opera 9.20 and, I'm told, IE7 all put the space node as a descendant of formatting elements as opposed to putting it as a child of body |
| 19:11 | <hsivonen> | jgraham: shouldn't this be raised with Anne first and with Hixie/the list only if Anne has a good reason to diverge from the spec and browser behavior? |
| 19:11 | <jgraham> | Yeah, that was expressed in reverse order |
| 19:16 | <hsivonen> | jgraham: email sent to Anne with you in the CC |
| 19:21 | <jgraham> | hsivonen: Got it. |
| 20:54 | <jgraham> | Philip` / hsivonen: Going back to the tokenizer tests for non-BMP entities; these (if I am not mistaken) assume the tokenizer produces UTF-16 output. |
| 20:55 | <jgraham> | Is that actually required anywhere? The DOM has to be UTF-16 but the HTML5 spec doesn't actually require that you construct a DOM per-se |
| 20:55 | <Philip`> | jgraham: They assume it produces UTF-32 output and then the JSON serialisation encodes those values as surrogate pairs |
| 20:56 | <hsivonen> | jgraham: no, JSON assumes UTF-16 code unit escapes |
| 20:56 | <hsivonen> | jgraham: if simplejson produces UTF-32 escapes, it is borked |
| 20:58 | <jgraham> | OK I think I understand. |
