| 00:08 | Philip` | wishes Opera had uneval |
| 00:18 | <Philip`> | Argh, rotateX uses degrees :-( |
| 00:19 | <Philip`> | Hooray, now I have a monkey and a box in Opera which looks about the same as in Firefox, except it's totally unshaded |
| 00:49 | <Philip`> | Now my monkey is shaded approximately the same in both, except rubbisher and slower in Opera |
| 00:49 | <Philip`> | but at least it works, which is handy |
| 01:19 | <Philip`> | http://philip.html5.org/demos/canvas/3d/x3d/test.html - should work in Firefox 3 beta with Canvas 3D extension and OpenGL 2.0 hardware; and in Opera 9.5 builds with opera-3d |
| 01:26 | <Philip`> | http://philip.html5.org/demos/canvas/3d/x3d/output-ff.png vs http://philip.html5.org/demos/canvas/3d/x3d/output-opera.png |
| 01:27 | <Philip`> | (The Opera version is several times slower, and has much more hardcoded lighting stuff) |
| 01:27 | <Philip`> | (and only does per-vertex lighting, not per-pixel) |
| 01:27 | <hdh> | wow, blender's monkey head |
| 01:27 | <Philip`> | Indeed :-) |
| 06:48 | <jacobolus> | Hixie: around? |
| 06:48 | <jacobolus> | I notice that the server-sent events things now have a big "might remove this" notice. does that mean altered, or scrapped altogether? |
| 06:49 | <Hixie> | either |
| 06:49 | <Hixie> | they're kinda redundant with xhr with multipart, and with the bidirectional networking stuff |
| 06:49 | <jacobolus> | Hixie: you see http://cometdaily.com/2007/12/11/the-future-of-comet-part-1-comet-today/ ? |
| 06:50 | <jacobolus> | Hixie: I have most of an article about server-sent events as well, talking about opera's implementation |
| 06:50 | <jacobolus> | also, apparently there are issues w/ multipart xhr |
| 06:50 | <jacobolus> | namely, there's no notice when connections are severed |
| 06:51 | <Hixie> | yeah we'd have to make sure we covered all the bases |
| 06:51 | <Hixie> | if we don't remove event-source, we'll probably simplify it a lot |
| 06:51 | <Hixie> | right now it's way overengineered |
| 06:53 | <jacobolus> | Hixie: yeah, that's a big part of my article :) |
| 06:53 | <jacobolus> | which is mostly actually about getting more ppl to send you feedback :) |
| 06:54 | <Hixie> | :-) |
| 06:54 | <Hixie> | to be honest even if we really want event-source and even if it was perfect, we might still have to remove them |
| 06:54 | <Hixie> | depends on whether browsers implement it or not |
| 06:54 | <jacobolus> | Hixie: if you have time to read this, let me know what you think :) (assuming the thing ever pastes) |
| 06:54 | <Hixie> | that goes for every feature of course |
| 06:54 | <Hixie> | the article? |
| 06:54 | <Hixie> | i read it |
| 06:55 | <jacobolus> | yeah, trying to paste to a pastebin |
| 06:55 | <jacobolus> | it's stalling |
| 06:55 | jacobolus | pasted http://pastie.textmate.org/private/xsvmjyopvftavgbcbjpexa |
| 06:55 | <jacobolus> | no no, the draft of the next one |
| 06:56 | <jacobolus> | Hixie: ↑ |
| 06:56 | Hixie | finishes reading it |
| 07:03 | <jacobolus> | hsivonen: I did a web search for "java applet suck", and a comment you made in here came up in the top page of google results :) |
| 07:06 | <jacobolus> | bb5 |
| 07:20 | <jacobolus> | Hixie: thoughts? |
| 07:22 | <jacobolus1> | Hixie: hrm, i was apparently disconnected. did i miss anything? |
| 07:36 | <jacobolus> | well. off to sleep. bb later sometime :) |
| 09:17 | <othermaciej> | Hixie: people have started on event-source for WebKit, so it would be good to have revision/simplification sooner rather than later |
| 09:17 | <othermaciej> | Hixie: I do tend to agree that sending arbitrary server-sent events to arbitrary nodes seems like overkill |
| 09:51 | <MikeSmith> | hsivonen - you around? |
| 09:52 | <MikeSmith> | I thought I remembered hsivonen saying that he had added some support to html5.validator.nu for checking for allowed values for the rel attribute based on scraping http://wiki.whatwg.org/wiki/RelExtensions |
| 09:58 | <hsivonen> | MikeSmith: I didn't add it yet. just discussed options with Hixie |
| 09:58 | <MikeSmith> | OK |
| 09:59 | <MikeSmith> | hsivonen - so html5.validator.nu isn't doing an conformance checking on rel values now, right? |
| 09:59 | <hsivonen> | MikeSmith: right |
| 09:59 | <MikeSmith> | OK, thanks |
| 09:59 | <hsivonen> | afk. switchin locations. will be back in 45 minutes |
| 10:03 | <hsivonen> | back here. bus gone already |
| 15:32 | <zcorpan_lap> | hsivonen: the group messages feature is great :) |
| 15:33 | <hsivonen> | zcorpan_lap: thanks |
| 15:34 | <hsivonen> | zcorpan_lap: re: <semantics>: annotation/annotation-xml are so under-defined that I seriously doubt that there are useful implementations |
| 15:35 | <zcorpan_lap> | hsivonen: ok |
| 15:36 | <hsivonen> | I think I'm going to leave annotation-xml as an anything-goes black hole for now |
| 15:36 | <hsivonen> | like RDF... |
| 15:38 | <zcorpan_lap> | hsivonen: http://validator.nu/?doc=http%3A%2F%2Fwww.w3.org%2F&schema=http%3A%2F%2Fsyntax.whattf.org%2Frelaxng%2Fhtml5full.rnc+http%3A%2F%2Fsyntax.whattf.org%2Frelaxng%2Fassertions.sch+http%3A%2F%2Fhsivonen.iki.fi%2Fcheckers%2Fall%2F&showsource=yes -- the location link for error 2 doesn't point to the right place |
| 15:38 | <zcorpan_lap> | s/error/message/ |
| 15:39 | <hsivonen> | Hixie: IIRC you said that RDF is allowed where metadata elements are allowed. are there other foreign metadata elements that I should allow in XHTML5? |
| 15:40 | <hsivonen> | zcorpan_lap: hmm. strange. I have a parser fix in my tree that I couldn't deploy yesterday. I try deploying now |
| 15:41 | <zcorpan_lap> | hsivonen: the link points to cl2c109 but the id is l2c109 |
| 15:41 | <hsivonen> | ooh. a front-end bug |
| 15:46 | <hsivonen> | zcorpan_lap: fixed. thanks |
| 15:46 | <hsivonen> | (deployed too) |
| 15:51 | <hsivonen> | can rdf:RDF legitimately have non-xmlns attributes? |
| 15:52 | <hsivonen> | Hixie: should only rdf:RDF be allowed as an RDF subtree root or should rdf:Description be allowed as well? |
| 15:53 | <zcorpan_lap> | hsivonen: ie doesn't know about HTMLOListElement |
| 15:54 | <hsivonen> | zcorpan_lap: ok. what test would you suggest instead? |
| 15:54 | <hsivonen> | I have a hard time interpreting http://www.w3.org/TR/rdf-syntax-grammar/#start |
| 15:55 | <zcorpan_lap> | hsivonen: .tagName.toLowerCase() == "ol" |
| 15:55 | <hsivonen> | zcorpan_lap: not theoretically pure, but OK |
| 15:56 | <zcorpan_lap> | hsivonen: indeed. not sure what else would work cross-browser |
| 15:57 | <hsivonen> | zcorpan_lap: does IE support nodeName? it would be more robust in case I ever put whitespace there |
| 15:58 | <zcorpan_lap> | hsivonen: it does |
| 15:58 | <zcorpan_lap> | hsivonen: you could check .start |
| 15:58 | <zcorpan_lap> | returns 1 if <ol> otherwise undefined |
| 15:59 | <zcorpan_lap> | http://software.hixie.ch/utilities/js/live-dom-viewer/?%3C!DOCTYPE%20html%3E%0D%0A%3Col%3E.%3Cscript%3Ew(document.body.firstChild.start)%3C%2Fscript%3E |
| 15:59 | <hsivonen> | hmm. I'm not sure which test is theoretically less pure... |
| 16:00 | <zcorpan_lap> | firefox returns -1 though |
| 16:02 | <hsivonen> | zcorpan_lap: yeah, I just noticed it broke firefox |
| 16:02 | <zcorpan_lap> | but -1 is truthy so you could just do if (n.start) |
| 16:02 | <hsivonen> | I did if (n.start) and it broke |
| 16:04 | <zcorpan_lap> | weird |
| 16:04 | <hsivonen> | hmm. file copying was what broke |
| 16:04 | <zcorpan_lap> | ok |
| 16:04 | <hsivonen> | now the script is on the server for real |
| 16:04 | <hsivonen> | zcorpan_lap: thanks |
| 16:05 | <zcorpan_lap> | np |
| 16:05 | <hsivonen> | XHTML in Safari still broken |
| 16:06 | <zcorpan_lap> | i think that's document.body |
| 16:06 | <zcorpan_lap> | hmm, now ie7 crashes |
| 16:06 | <zcorpan_lap> | :) |
| 16:13 | <hsivonen> | zcorpan_lap: any advice on what I need to do to stop it from crashing? |
| 16:13 | <zcorpan_lap> | i don't know what it is that makes it crash |
| 16:14 | <zcorpan_lap> | it's not the .start check, i think |
| 16:15 | <hsivonen> | I think I'm doing relatively tame DOM operations |
| 16:15 | <zcorpan_lap> | it's not createHtmlElement either |
| 16:18 | <zcorpan_lap> | hsivonen: you may need to set .type = 'button' before you append the input to the p |
| 16:19 | <zcorpan_lap> | http://software.hixie.ch/utilities/js/live-dom-viewer/?%3C!DOCTYPE%20html%3E%0D%0A%3Cscript%3E%20function%20createHtmlElement(tagName)%7B%0D%0A%20%20%20%20return%20document.createElementNS%20%3F%20document.createElementNS(%22http%3A%2F%2Fwww.w3.org%2F1999%2Fxhtml%22%2C%20tagName)%20%3A%20document.createElement(tagName)%0D%0A%7D%0D%0Avar%20currentOl%20%3D%20document.body.firstChild%3B%0D%0Afunction%20installGrouping |
| 16:19 | <zcorpan_lap> | a%20%3D%20createHtmlElement('p')%0D%0A%09var%20button%20%3D%20createHtmlElement('input')%0D%0A%09para.appendChild(button)%0D%0A%09button.type%20%3D%20'button'%0D%0A%7D%0D%0AinstallGroupingToggle()%0D%0A%3C%2Fscript%3E |
| 16:20 | <hsivonen> | zcorpan_lap: better now (after reload)? |
| 16:21 | <zcorpan_lap> | still crashes |
| 16:22 | <hsivonen> | zcorpan_lap: you were right about document.body and Safari |
| 16:28 | zcorpan_lap | has no idea what makes ie crash :( |
| 16:29 | <hsivonen> | zcorpan_lap: does it crash onload? |
| 16:29 | <hsivonen> | that is, before pushing the button? |
| 16:29 | <zcorpan_lap> | yes |
| 16:29 | <zcorpan_lap> | but not before you've validated something |
| 16:30 | <zcorpan_lap> | i.e. just loading http://validator.nu/ works fine |
| 16:30 | <hsivonen> | well, crashing isn't user-friendly |
| 16:30 | <zcorpan_lap> | :) |
| 16:30 | <hsivonen> | I guess it's better if I just disable the script for IE |
| 16:31 | <hsivonen> | I wonder what's the best way to *really* detect IE without depriving other browsers of the script |
| 16:31 | <hsivonen> | should I just put the instanceof thing back? |
| 16:32 | <zcorpan_lap> | that would work |
| 16:32 | <zcorpan_lap> | the safest way otherwise, i think, is conditional comments (either in html or js) |
| 16:35 | <hsivonen> | zcorpan_lap: I put instanceof back |
| 16:35 | <Philip`> | Can you report the bug to the IE developers? |
| 16:35 | <hsivonen> | Philip`: me? I haven't even witnessed it firsthand |
| 16:37 | <Philip`> | hsivonen: I guess "you" means "someone who can reproduce the bug" |
| 16:37 | Philip` | doesn't know if there's still any public bug reporting mechanism |
| 16:38 | <zcorpan_lap> | "Now that the final version of IE7 has been released, the IE Feedback site is temporarily closed. It will re-open in the future." |
| 16:38 | <zcorpan_lap> | it has been "temporarily closed" the past year or so |
| 16:39 | <hsivonen> | did anne write an XBL2 schema in RELAX NG? |
| 16:40 | hsivonen | works on a massive überschema reorganization |
| 16:51 | hsivonen | has trouble getting trang to cooperate... |
| 17:23 | <hasather> | hsivonen: the schemas are here: http://dev.w3.org/cvsweb/2006/waf/xbl-schemas/ |
| 17:23 | <hsivonen> | hasather: thanks |
| 17:27 | <hsivonen> | hmm. anne is not here. that xbl2 schema lacks license info |
| 17:30 | <hsivonen> | has Hixie really specced XBL id to take the value constraints of xsd:ID or has Anne taken liberties with the schema? |
| 17:34 | <hasather> | hsivonen: as far as I know, that was never specced |
| 17:35 | <hsivonen> | hasather: OK |
| 17:37 | <hasather> | hsivonen: I may be wrong though, but I remember there were some discussion around ID in XBL |
| 17:38 | <hsivonen> | hasather: there were at least the discussions where nwalsh said it should use xml:id |
| 17:38 | <hasather> | yea |
| 17:49 | <gsnedders> | Old glow sticks are so boring. |
| 18:47 | <jruderman> | gsnedders: break it open and pour it in homestar's mountain dew |
| 19:26 | <hsivonen> | hmm. the message grouping script has a silly bug when source is shown and there are no messages... |
| 19:34 | <zcorpan_lap> | hsivonen: i guess you could just set an id on the messages list |
| 19:34 | <zcorpan_lap> | although if you replace instanceof with getElementById then you'll probably get the ie crasher again :) |
| 19:35 | <hsivonen> | zcorpan_lap: I wanted to avoid issues like that :-) |
| 19:40 | <hsivonen> | let me say again that RELAX NG DTD Compatibility sucks |
| 19:47 | <hsivonen> | w00t. my local copy validates intertwingly.net/blog |
| 19:47 | <hdh> | what can I use instead of XFN? |
| 19:48 | <hsivonen> | hdh: for what purpose? |
| 19:48 | <hdh> | in html5 the rel values is a limited set → no XFN |
| 19:48 | <hsivonen> | hdh: please edit the wiki :-) |
| 19:52 | <hdh> | "Unless otherwise specified, a keyword must not be specified more than once per rel attribute." ← "keyword" here means link type? |
| 19:53 | <hsivonen> | yes |
| 20:08 | <hdh> | hm, unknown rel types aren't conformance error, and the spec does say that "if none of the values in the attribute are recognised by the UA", i.e. allowing types outside it |
| 20:08 | <hdh> | s/that// |
| 20:13 | <Philip`> | hdh: 4.12.3.21 says "Other than the types defined above, only types defined as extensions in the WHATWG Wiki RelExtensions page may be used with the rel attribute on link, a, and area elements." |
| 20:14 | <Philip`> | so it sounds like it's intending that unknown types (i.e. not in spec nor wiki) are an error |
| 20:14 | <Philip`> | though it should probably say "must" somewhere, and I can't see that it does |
| 20:16 | <hsivonen> | any opinions on what I should do with inkscape/sodipodi product-specific cruft? |
| 20:17 | <hdh> | search for prefix:, no hit → delete xmlns: ; remove all style= and test if things look ugly afterwards; run throught xmllint to drop whitespace |
| 20:17 | <hdh> | I haven't tried the second one yet |
| 20:18 | <hsivonen> | hdh: I meant: what should the validator do |
| 20:18 | <Philip`> | hsivonen: It would be non-helpful if it was an error, since there's no point in authors cleaning it up |
| 20:18 | <Philip`> | (as far as I'm aware) |
| 20:18 | <hdh> | warn as unknown namespace? |
| 20:18 | <hsivonen> | Philip`: will I be product-biased if I allow sodipodi/inkscape cruft specifically? |
| 20:19 | <Philip`> | though maybe it could have one warning saying "you're wasting some space with this stuff - try the 'export as plain SVG' option" |
| 20:19 | <hsivonen> | or of course I would, but would I be bad? |
| 20:20 | <Philip`> | hsivonen: I don't see why it would be bad, as long as you gave the same allowance for any other product that people use and which is harmless |
| 20:20 | <Philip`> | (where "that people use" could mean "that people complain to you about") |
| 20:21 | <hsivonen> | Philip`: ok |
| 20:21 | <Philip`> | I could be wrong, of course :-) |
| 20:21 | hsivonen | wonders if that's a slippery slope to allowing RDFa in XHTML5 :-) |
| 20:22 | <Philip`> | Hmm, can you validate RDF that's inside HTML comments? |
| 20:22 | <hsivonen> | Philip`: I don't |
| 20:23 | <hsivonen> | Philip`: however, I'm right now punching holes for real RDF in the XHTML5 and SVG 1.1 schemas |
| 20:26 | Philip` | wonders if anyone would want to embed X3D inline in XHTML |
| 20:26 | <Philip`> | (That would be easier if it had a namespace, I suppose) |
| 20:27 | <Philip`> | (but it's easy to make one up) |
| 20:28 | <zcorpan_lap> | hsivonen: it seems weird to allow sodipodi/inkscape cruft if it's not allowed per the svg spec |
| 20:30 | <Philip`> | "SVG allows inclusion of elements from foreign namespaces anywhere with the SVG content. In general, the SVG user agent will include the unknown elements in the DOM but will otherwise ignore unknown elements." |
| 20:30 | <Philip`> | "Additionally, SVG allows inclusion of attributes from foreign namespaces on any SVG element. The SVG user agent will include unknown attributes in the DOM but with otherwise ignore unknown attributes." |
| 20:30 | <Philip`> | That sounds like it's allowed |
| 20:30 | <Philip`> | (via http://www.w3.org/TR/SVG11/extend.html ) |
| 20:31 | <zcorpan_lap> | ok. then you wouldn't have to have knowledge about sodipodi/inkscape cruft specifically |
| 20:32 | <Philip`> | You would have to if you want to whitelist known extensions, since otherwise you couldn't detect people incorrectly using xmlns="http://www.w3.org/1999/html"; or whatever |
| 20:32 | <hsivonen> | I'm not a fan of anything-goes holes |
| 20:32 | <hsivonen> | Philip`: indeed |
| 20:33 | <hsivonen> | btw, the above "ignore" text doesn't say if ignoring an element means ignoring the subtree or behaving as if the content of the element appeared in place of the element |
| 20:34 | <hsivonen> | I'm inclined to restrict foreignObject to XHTML and MathML only |
| 20:35 | <hdh> | is xfn not allowed on <link>? |
| 20:35 | <hsivonen> | hdh: it is if it's on the wiki. if not, it will be once you've edited the wiki |
| 20:35 | <hdh> | I'm editing it, but not sure what to put in the Effect on link column |
| 20:36 | <hsivonen> | "(Other XML-compatible metadata languages, including ones not based on RDF, can be used also.)" Any examples of "other"? |
| 20:36 | <zcorpan_lap> | is there a spec for xfn? |
| 20:36 | <hdh> | http://gmpg.org/xfn/11 |
| 20:36 | <hsivonen> | http://gmpg.org/xfn/11 |
| 20:37 | <hsivonen> | I'm slow |
| 20:37 | <zcorpan_lap> | thanks |
| 20:37 | <hdh> | actually a profile, which is gone in html5 |
| 20:38 | <hsivonen> | doesn't matter |
| 20:38 | <hsivonen> | no one checks the profile attribute upon processing |
| 20:39 | <Philip`> | I thought GRDDL does |
| 20:40 | <Philip`> | (since the profile points to some XSLT to convert the XHTML to RDF, according to this diagram) |
| 20:40 | <hsivonen> | ok, someone checks |
| 20:40 | <hsivonen> | except the XFN profile isn't XSLT. oops. |
| 20:41 | <hdh> | and xslt is a stylesheet |
| 20:41 | <hsivonen> | a "transformation" would be more politically correct on this channel :-) |
| 20:42 | <Philip`> | http://www.w3.org/2003/12/rdf-in-xhtml-xslts/grokXFN.xsl |
| 20:42 | <hdh> | aye, I'm a newcomer |
| 20:43 | <hsivonen> | Philip`: is there some mapping magic from a gmpg.org URI to that URI? |
| 20:45 | <Philip`> | hsivonen: Don't know - I'm not sure if the aim is to put the XSLT version at the gmpg.org URI, or to have an external mapping that doesn't need cooperation, or something else |
| 20:46 | <Philip`> | I vaguely remembering hearing something about those possibilities, except I've forgotten it all now |
| 20:46 | <Philip`> | s/remembering/remember/ |
| 20:46 | <hsivonen> | Does illustrator put any private-namespace cruft in SVG? |
| 20:47 | <gsnedders> | hsivonen: no |
| 20:47 | <hsivonen> | gsnedders: thanks |
| 20:47 | <gsnedders> | (at least, I've never seen so in CS3) |
| 20:50 | <Philip`> | The examples at http://www.adobe.com/svg/demos/samples.html don't have any public namespaces (and aren't well-formed) either :-/ |
| 20:50 | <Philip`> | and have Content-Type: text/plain |
| 20:51 | <hsivonen> | XML is hard. |
| 20:52 | <Philip`> | (Doesn't look like those samples come from Illustrator anyway) |
| 20:52 | <hsivonen> | I'd like to see real-world cases of RDF inside SVG inside XHTML |
| 20:54 | <Philip`> | http://www.adobe.com/svg/svgfiles/interactive_filters.svgz |
| 20:54 | <gsnedders> | Philip`: how come on http://tinyurl.com/2dsa9f line 105 it is returned before it is set on line 98? Am I getting something wrong with how variables bounce around from function to function? |
| 20:54 | <Philip`> | is apparently from Illustrator |
| 20:54 | <Philip`> | <svg xmlns:i="&ns_ai;" xmlns:xlink="&ns_xlink;" xmlns:a="http://ns.adobe.com/AdobeSVGViewerExtensions/3.0/"; width="400" height="304.9" viewBox="0 0 400 304.9" style="overflow:visible;enable-background:new 0 0 400 304.9" xml:space="preserve"> |
| 20:54 | <Philip`> | and <!ENTITY ns_ai "http://ns.adobe.com/AdobeIllustrator/10.0/">; etc |
| 20:54 | <hdh> | er, the second newest blog post from Shelley |
| 20:55 | <hsivonen> | hdh: thanks |
| 20:56 | <Philip`> | The <svg> is still in no namespace, but at least Firefox displays the XML tree instead of a parse error |
| 20:56 | hdh | 's firehose has been planet intertwingly lately |
| 20:56 | Philip` | wonders if Adobe cares about anything other than the Adobe SVG Viewer |
| 20:56 | <hsivonen> | Philip`: eww. private elements |
| 20:56 | <hsivonen> | no fun |
| 20:57 | gsnedders | opens Ai CS3 |
| 20:58 | <Philip`> | gsnedders: XMLHttpRequest is asynchronous, so .send() returns before it's sent the request and called onreadystatechange and set ok |
| 20:58 | <zcorpan_lap> | hmm |
| 20:59 | <gsnedders> | hsivonen: CS3 has SVG and Xlink namespaces only, no DOCTYPE |
| 20:59 | <hsivonen> | hmm. Shelley relies on entities in XML on the Web... |
| 20:59 | <zcorpan_lap> | if illustrator doesn't emit xmlns="...svg", then it wouldn't be helpful to require xmlns="...svg" for text/html |
| 21:00 | <gsnedders> | Philip`: will setting it to false break anything? |
| 21:00 | <Philip`> | gsnedders: Setting what to false? |
| 21:00 | <gsnedders> | async |
| 21:02 | <Philip`> | gsnedders: Hmm, I guess it'd probably work if you change both requests to be synchronous, though you might end up with complaints about scripts taking too long to run |
| 21:03 | <hsivonen> | SVG element metadata not allowed as child of SVG element svg in this context. |
| 21:03 | <hsivonen> | I wonder why that happens... |
| 21:03 | <hsivonen> | Bad value 1.0 for attribute version on SVG element svg. |
| 21:03 | <hsivonen> | that's useful... Not |
| 21:04 | <hsivonen> | "The first occurrance of the ID value defs3 was here." hazards of mere inclusion... |
| 21:05 | <Philip`> | s/occurrance/occurrence/ |
| 21:08 | <Philip`> | From http://burningbird.net/technology/standards-take-work/ : "sodipodi:docbase="/Users/richardthompson/Desktop"" - that looks like unintentional information leakage |
| 21:08 | <Dashiva> | Say hsivonen, would the validator be able to validate DOM trees too (in a suitable serialization format, like the one zcorpan was working on)? |
| 21:08 | <hsivonen> | Philip`: thanks. now fixed in my local copy (not deployed) |
| 21:09 | <hsivonen> | Dashiva: if you contribute a parser for zcorpan's format to SAX |
| 21:10 | <Philip`> | Dashiva: (Why not a suitable serialisation format like XML?) |
| 21:10 | <Dashiva> | Because not all DOM trees can be represented as XML |
| 21:10 | <Philip`> | Ah |
| 21:10 | <hsivonen> | Dashiva: it would be nice to have a Firefox extension that'd take a snapshot of the DOM and would send in to Validator.nu |
| 21:11 | <Dashiva> | Yeah, I got the idea from opera's validation option, and thought to extend it to a page after e.g. onload scripts had done their magic |
| 21:11 | <Philip`> | It would be nicer to have one that does local validation, so it doesn't send e.g. your filled-in password and credit card number fields to Validator.nu |
| 21:12 | <Dashiva> | But how big is the validator, with all dependencies? :) |
| 21:12 | <Philip`> | Do you count the JVM as a dependency? :-) |
| 21:12 | <hsivonen> | Dashiva: it's big |
| 21:13 | <hsivonen> | and there's the JVM startup delay |
| 21:13 | <Philip`> | Hopefully you would only have the startup delay when launching Firefox the first time, and ideally that would happen asynchronously |
| 21:14 | <Philip`> | s/the first time// |
| 21:16 | hdh | just noticed the RDF tags for Shelley's tree in akregator |
| 21:16 | <Dashiva> | Might be better to simply run a pre-hook that removes value attributes from password fields :) |
| 21:17 | <hsivonen> | Dashiva: if you are OK with non-XMLizable DOM proclaimed invalid, the current Web service API goes a long way with XHR |
| 21:17 | <Philip`> | After you've sent your hundred-kilobyte document the validator a few dozen times, you'd be better off in terms of bandwidth if you downloaded a local copy plus all the dependencies |
| 21:17 | <hsivonen> | btw, what does XHR do if you try to POST a tree that doesn't serialize to XML? |
| 21:18 | <Philip`> | (particularly since developers will run this from their own computers, so it's not transmitting from the web server, so it'll be a slower network connection) |
| 21:19 | <hsivonen> | If someone wants to write a .zip packaging action for build/build.py, I'd be happy to integrate it |
| 21:23 | <zcorpan_lap> | setting opera:config#UserPrefs|ValidationURL to http://validator.nu/ seems to work |
| 21:23 | <hsivonen> | zcorpan_lap: cool |
| 21:23 | <hdh> | very |
| 21:23 | <hsivonen> | unintentional compatibility |
| 21:24 | <zcorpan_lap> | although encoding information is lost |
| 21:24 | <zcorpan_lap> | hsivonen: it would be useful to be able to suggest and/or override encoding |
| 21:25 | <hsivonen> | zcorpan_lap: on file upload? |
| 21:25 | <hsivonen> | zcorpan_lap: or in general_ |
| 21:25 | <hsivonen> | ? |
| 21:25 | <zcorpan_lap> | at least file upload |
| 21:26 | <hsivonen> | zcorpan_lap: feature request recorded. thanks |
| 21:26 | <hdh> | web developer "validate local html" doesn't have uri setting, but the other one do |
| 21:28 | zcorpan_lap | tried to enter http://validator.nu/?showsource=yes in opera:config but it didn't have the desired effect :( |
| 21:30 | <hsivonen> | zcorpan_lap: the query string is ignored in the file upload and textarea modes |
| 21:31 | <zcorpan_lap> | hsivonen: why? |
| 21:34 | <hsivonen> | zcorpan_lap: there's different code that reads params from multipart/form-data |
| 21:34 | <hsivonen> | zcorpan_lap: recorded as a feature request |
| 21:34 | <zcorpan_lap> | hsivonen: ok. thanks |
| 21:46 | <gsnedders> | Philip`: you're original only ever could be waiting on one thing at a time anyway |
| 21:47 | <gsnedders> | Philip`: making it synchronous would have no difference from your original |
| 21:47 | <Philip`> | From X3D / H-Anim: "float[4] rotation 0 0 1 0" ... "The rotation field specifies a rotation of the coordinate system of the Joint object as a quaternion." - it would be so much better if they actually defined quaternions and how they're represented, rather than making you guess whether it's xyzw or wxyz |
| 21:48 | <Philip`> | gsnedders: The original asynchronous one lets the browser get on with other things in the meantime, whereas the synchronous one prevents the browser from doing anything much to the page while the script is running and waiting |
| 21:48 | <gsnedders> | Philip`: yeah, I know. But how the original version of the script is written, it never tries to load more than one file at once |
| 21:48 | <othermaciej> | Philip`: quaternion! crazy |
| 21:49 | <Philip`> | gsnedders: The server is single-threaded so it can only load one file at a time anyway :-) |
| 21:49 | <gsnedders> | Philip`: the browser get on with what other things? surely other things outwith JS are unaffected by it? |
| 21:50 | <gsnedders> | (forgive my naïvety about this) |
| 21:50 | Philip` | actually likes quaternions |
| 21:51 | <Philip`> | gsnedders: Maybe things like noticing DOM changes (e.g. when log messages are added) and re-rendering, except I don't know which browsers get hung up on that in practice |
| 21:52 | <hsivonen> | I was excited about the RNC "external" keyword when I learned about it but now I'm disappointed because I don't know how to make circular references |
| 21:52 | <gsnedders> | Philip`: but if the script isn't running, we don't add log mesages |
| 21:53 | gsnedders | wonders whether to make the log an ordered list |
| 21:53 | <Philip`> | gsnedders: The script is 'running' in the sense that it's in the middle of executing a JS function, while waiting for the synchronous XHR, and it never stops running until every test has been processed |
| 21:53 | <gsnedders> | Philip`: I mean, the script isn't moving forward. |
| 21:53 | <gsnedders> | Philip`: so therefore we don't add log messages, we're just waiting at one line |
| 21:54 | <gsnedders> | Philip`: but you think that browsers don't actually check the DOM till JS has totally finished executing? :\ |
| 21:55 | <hsivonen> | is SVG 1.1 Full a superset of SVG 1.0 and SVG 1.2 Full a superset of SVG 1.1 Full? |
| 21:55 | <Philip`> | gsnedders: It does (I assume) something effectively like "function run_tests() { log('Test 1'); xhr_request_stuff(); log('Test 2'); xhr_request_stuff(); ... }" and it doesn't return until everything's finished, so the browser doesn't get a chance to dispatch events (since it has to pretend the script is running single-threadedly) |
| 21:55 | <Philip`> | ...I think |
| 21:56 | <Philip`> | gsnedders: I think Opera and Firefox vary in whether/when they redraw the screen in response to DOM changes if a script is still running, but I can't remember the details now |
| 21:56 | <gsnedders> | Silly browsers. |
| 21:56 | <Philip`> | They'd be sillier if they allowed arbitrary concurrency between scripts :-) |
| 21:57 | <hsivonen> | any opinions on what I should do with <svg version='1.0'>? complaining about "1.0" seem useless |
| 21:58 | <hsivonen> | but then silently accepting version='1.0' with 1.1 features seems bogus, too |
| 21:58 | <gsnedders> | Philip`: events in the readystatechange event sense, or? |
| 21:58 | <Philip`> | hsivonen: Use a SVG 1.0 schema? |
| 21:59 | <Philip`> | gsnedders: Yes, but I guess readystatechange in synchronous XHR is special |
| 22:00 | <hsivonen> | Philip`: is there one that a) I don't need to write, b) has reasonable quality and c) doesn't conflict with the production names from the 1.1 schema? |
| 22:00 | <Philip`> | hsivonen: No idea. Those do sound like possible difficulties |
| 22:00 | <hsivonen> | point c) is particularly noteworthy if I'm expected to support 7 flavors of SVG |
| 22:01 | <Philip`> | Can't you have namespaced schemas? :-) |
| 22:02 | <hsivonen> | Philip`: I could, but it's a non-trivial find-and-replace job |
| 22:02 | <hsivonen> | or I could do that step in RNG and only then convert to RNC |
| 22:03 | <hsivonen> | what I really want is an SVG5 that gives me a license to treat version='' as a talisman |
| 22:08 | <othermaciej> | does SVG actually require particular treatment of version? |
| 22:08 | <Philip`> | X3D says "a script url prefixed with ecmascript: (or the deprecated javascript:) shall contain ECMAScript source" - is ecmascript: used by anything else? |
| 22:08 | <othermaciej> | no |
| 22:08 | <hsivonen> | othermaciej: I don't know |
| 22:08 | <Philip`> | I wish they gave examples in the XML syntax rather than the VRML syntax |
| 22:09 | <hsivonen> | so far, it seems to me that version='' only causes purity issues for validators |
| 22:10 | Philip` | wonders if he should bother trying to implement a subset of X3D timing/events/scripting, or just make up something easier to do animation |
| 22:20 | <gsnedders> | How bizarre. 'Got "connection" header "", expected "close"' — Op9.2 |
| 22:25 | <Philip`> | http://www.web3d.org/specifications/x3d-3.0.xsd - hmm, text/plain |
| 22:25 | <Philip`> | Actually, 'text/plain; charset=iso-8859-1' and then <?xml version="1.0" encoding="UTF-8"?> |
| 23:19 | <Philip`> | Hmm, an X3D spec example uses a TimeSensor field named "fraction", which actually doesn't exist in the definition of TimeSensor (whereas there is a "fraction_changed" field), but it works anyway in at least one particular X3D viewer |
| 23:20 | Philip` | won't support that bug for now |
