| 00:03 | <Lachy> | hey Hixie |
| 00:03 | <Lachy> | did you see my previous message about acid3? |
| 00:03 | <Lachy> | from about 12 hours ago |
| 00:08 | <Hixie> | acid3.acidtests.org/acid3-2008-03-28.tar.gz |
| 00:08 | <Hixie> | iirc |
| 00:09 | <Lachy> | thanks |
| 00:13 | <Lachy> | hmm. strange. When I run that locally, tests 4 and 5 fail in webkit |
| 00:15 | dglazkov | watches Hixie carefully so that he doesn't become evil |
| 00:15 | <Hixie> | hehe |
| 00:15 | <dglazkov> | that was a little bit evil |
| 00:15 | <Hixie> | :-/ |
| 00:16 | <dglazkov> | :) |
| 00:16 | <annevk> | there was some amount of laughter here when glazou read "this won't really change anything;" |
| 00:16 | <dglazkov> | hey, annevk, do you need webforms2.org? |
| 00:17 | <annevk> | oh, right, did you e-mail about that? |
| 00:17 | <annevk> | i wasn't really sure what to do with it |
| 00:17 | <annevk> | apart from paying 10 bucks a year |
| 00:17 | <dglazkov> | me neither :) |
| 00:18 | <annevk> | also, at some point it'll just be HTML5 Forms |
| 00:18 | <annevk> | or HTML forms |
| 00:19 | <dglazkov> | well, the whole HTML5 thing does kind of fade the newness and edginess of webforms2 |
| 00:19 | <dglazkov> | perhaps there should be a monument at webforms2.org -- here began the quest to better Web. |
| 00:19 | <annevk> | hehe |
| 00:19 | <annevk> | i suppose i'm willing to pay for that :) |
| 00:20 | <annevk> | if you e-mail me the transfer code i can do the rest, i have to go now though |
| 00:20 | <Lachy> | dglazkov, stick some ads on it. That's what the spammers will do the instant you let the domain expire, so you may as well profit before they do :-) |
| 00:32 | <othermaciej> | Lachy: we have a gently tweaked local copy in our LayoutTests that passes |
| 01:39 | <Hixie> | aw man i can't wait for GTAIV to come out |
| 01:39 | <Hixie> | one month! |
| 02:14 | <Hixie> | woot! |
| 02:14 | <Hixie> | i used an <hr> on my last blog post |
| 02:14 | <Hixie> | and it doesn't look like crap! |
| 02:16 | <csarven> | Hixie Haven't seen a proper use of <hr> in a *long* time. |
| 02:16 | <Hixie> | yeah |
| 02:16 | <Hixie> | i think "semantic-thinking" people avoid it |
| 02:17 | <Hixie> | thinking it's "presentational" |
| 02:17 | <Hixie> | i did |
| 02:17 | <Hixie> | until changing html5 to make it clearly semantic |
| 02:17 | <Hixie> | :-) |
| 02:18 | <csarven> | Gar. Gotta run. |
| 02:18 | <csarven> | How did you get the asterisks going? |
| 02:19 | <csarven> | BBL. I'll check the channel logs. |
| 04:19 | <Hixie> | csarven: generated content |
| 05:27 | <Hixie> | so... any opinions on http://wiki.whatwg.org/wiki/New_Vocabularies ? |
| 06:14 | <Hixie> | anyone have a list of html elements? |
| 06:26 | Hixie | investigates mathml further |
| 07:52 | <virtuelv> | Hm, http://en.wikipedia.org/wiki/Acid_3 is out of date |
| 07:54 | <othermaciej_> | looks like someone edited out "This likely reduced Opera's score to 99/100" too |
| 08:20 | <virtuelv> | othermaciej: I was specifically refering to "By a changed of the Testsuit Opera will only achieve 99/100 now." |
| 08:22 | <othermaciej> | afaik (though I never saw the original internal build in action) the test suite change did lower the score, but it was fixed again before the release |
| 08:22 | <othermaciej> | (i.e. fixed again in GOGI) |
| 08:22 | <othermaciej> | nothing in the page claims 99/100 any more afaict |
| 08:31 | <hsivonen> | Hixie: the current extension point of MathML, annotation-xml, has the nice property that it could create a parsing scope without having to hard-code particular HTML elements as DOM namespace changing |
| 09:22 | <hsivonen> | hmm. I subscribed to public-pfwg-comments but I'm not getting email from the list. |
| 09:22 | <hsivonen> | instead I'm getting an autoreply each time I send something there |
| 09:31 | <Lachy> | hsivonen, mail sysreq⊙wo and ask them to fix it |
| 09:31 | <hsivonen> | Lachy: I'll do that. thanks |
| 10:30 | <hsivonen> | http://blogs.sun.com/bblfish/entry/rdfauth_sketch_of_a_buzzword |
| 11:03 | <jgraham__> | hsivonen: I had the same experience with pfwg-comments |
| 15:04 | <BenMillard> | PFWG is not sending me mail but I didn't get any autoreply for the message I sent them, either |
| 15:04 | <BenMillard> | here's what I did to subscribe: http://projectcerbera.com/blog/2008/03#day28 |
| 16:03 | <gsnedders> | Philip`: do you have HTTP header data for anything apart from the 15k dmoz dataset? |
| 16:05 | <Philip`> | gsnedders: I have it for a ~130K dmoz dataset too |
| 16:06 | <gsnedders> | Philip`: Could I possibly get that in the same XML grep format as before? |
| 16:07 | <gsnedders> | i.e., <header name="Date" uri="http://www.grebe-gt.de/"; value="Wed, 12 Dec 2007 00:19:58 GMT"/> within root xml_grep element |
| 16:08 | <Philip`> | gsnedders: I don't trivially have it in that format, but I have something very similar - should that be good enough? |
| 16:08 | <gsnedders> | Philip`: what sort of format? |
| 16:08 | <Philip`> | (The <header> bits are the same, but the root is <survey> and it has <processed> and <redirect> and <error> stuff) |
| 16:08 | <gsnedders> | send it over, then I'll bitch if I don't like it :) |
| 16:08 | <Philip`> | It's only a hundred megabytes of XML... |
| 16:09 | <gsnedders> | I was expecting that. |
| 16:09 | <gsnedders> | I'll work with it somehow :) |
| 16:09 | Philip` | wonders how best to compress it |
| 16:09 | <Philip`> | I want to minimise time taken to transfer across internet + time to compress + time to decide how to compress |
| 16:10 | <Philip`> | Oh, bzip2 was quite fast |
| 16:10 | <Philip`> | http://www.cl.cam.ac.uk/~pjt47/misc/output-headers.xml.bz2 (10MB) |
| 16:11 | <Philip`> | Oh |
| 16:11 | <Philip`> | Is something sending the wrong content-type for that? |
| 16:11 | <gsnedders> | yeah |
| 16:11 | <gsnedders> | it doesn't batter forme tho |
| 16:11 | <gsnedders> | *matter |
| 16:11 | <gsnedders> | *for me |
| 16:11 | <gsnedders> | *though |
| 16:11 | <gsnedders> | oh dear. |
| 16:12 | <Philip`> | Sadly the "BZh91AY&" at the front causes XML well-formedness errors |
| 16:12 | <Philip`> | ? |
| 16:13 | <Philip`> | Oh |
| 16:13 | Philip` | should try comprehending all of gsnedders' lines before failing to understand the last one |
| 16:13 | <Philip`> | Let me know once you've downloaded it, and I'll delete it again to free some disk space |
| 16:14 | <gsnedders> | Philip`: the oh dear was in reference to the number of mistakes |
| 16:14 | <Philip`> | I worked that out eventually :-) |
| 16:15 | gsnedders | checks it is well-formed |
| 16:15 | <gsnedders> | Philip`: sure, delete it |
| 16:16 | <Philip`> | (I can move it back if anyone else wants a copy) |
| 16:16 | <gsnedders> | (I'll make it public shortly anyway) |
| 16:19 | <gsnedders> | Philip`: how much of <http://hg.gsnedders.com/http-parsing/file/2f867e70fcfc/Philip%20Taylor%27s%20Header%20Data/README.txt>; is true for the new data? |
| 16:33 | <Philip`> | gsnedders: It's not ~15k any more |
| 16:34 | <gsnedders> | Philip`: OK, of the less than totally obvious ones :) |
| 16:34 | <gsnedders> | Philip`: second paragraph? parsed by HttpClient? |
| 16:35 | <Philip`> | gsnedders: The <error>s are given for non-200 responses, but I don't know if that matters |
| 16:35 | <gsnedders> | I'm writing a DTD for the format, FWIW, mainly to check what I think the format is is correct :) |
| 16:36 | <Philip`> | It's still HttpClient, though this time I changed the headers from the defaults to include |
| 16:36 | <Philip`> | Accept: text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8 |
| 16:36 | <Philip`> | Accept-Charset: ISO-8859-1,utf-8;q=0.7,*;q=0.7 |
| 16:36 | <Philip`> | Accept-Encoding: gzip,deflate |
| 16:36 | <Philip`> | Accept-Language: en-us,en;q=0.5 |
| 16:36 | <gsnedders> | from what before? |
| 16:37 | <Philip`> | From whatever HttpClient's default were |
| 16:37 | <Philip`> | (which I think didn't include any of those headers) |
| 16:37 | <Philip`> | (These are copied from some roughly-FF3b4 nightly) |
| 16:53 | <gsnedders> | Philip`: what's wrong with <!ELEMENT survey ((processed, redirect?, header+) | error)*>? |
| 16:55 | <Philip`> | gsnedders: Several things |
| 16:55 | <gsnedders> | :D |
| 16:56 | gsnedders | assures everyone here he really has got the format wrong |
| 16:56 | <Philip`> | 'error' gets output if there's a Java exception at some point during the processing, so it's quite possible that there'd be an <error> after it's started writing some of the other bits |
| 16:56 | <Philip`> | ((Also it gets output on non-200 responses etc - it's not just exceptions)) |
| 16:57 | <Philip`> | For each URI the elements are in about that order, but the output for each URI can be arbitrarily interleaved with the output from others |
| 16:58 | <Philip`> | so if you're ignoring attribute values, the element ordering is basically completely unstructure |
| 16:58 | <Philip`> | d |
| 17:00 | <gsnedders> | Philip`: if it is interleaved, how does processed mean anything? |
| 17:02 | <Philip`> | gsnedders: It doesn't really mean much - just that the page was downloaded 'successfully' |
| 17:02 | <Philip`> | Mostly it's useful so I can count how many pages there are |
| 17:02 | <Philip`> | (just using grep and wc) |
| 17:02 | Philip` | goes away for a while |
| 17:06 | <gsnedders> | nano seems to be the only text editor that is anything near performent on a file of that size |
| 17:37 | Philip` | uses 'less' |
| 19:32 | <Hixie> | man, the number of people who keep suggesting that we could add a mode in the parser for things like <annotation-xml> suggests that I'm going to have to give detailed explanations of the problems of doing that |
| 19:38 | <jgraham__> | Hixie: Did you really say that "you do not intend to allow SVG in text/html"? |
| 19:38 | jgraham__ | doesn't remember that |
| 19:39 | <Hixie> | i believe at one point while fixing the acid3 test i got so annoyed at the vagueness of the svg spec and the bad design of the svg language that i may have said something along the lines of "right, screw svg, i'm not putting it in html" |
| 19:39 | <Hixie> | i really wish there was a better option though |
| 19:39 | <jgraham__> | Ah, OK |
| 19:40 | <Hixie> | svg is just such a crappy language |
| 19:40 | <jgraham__> | I suggest you never look at the aria spec :) |
| 19:41 | <jgraham__> | I don't really know about whether SVG is crappy or not but it seems to me that, given the extant implementations, it's the sensible choice for vector graphics in text/html |
| 19:41 | <Hixie> | the svg spec has the quadruple disadvantages of being huge, being badly designed, being designed in a way to fit a mythical w3c architecture (e.g. it uses xlinks for no good reason), and being ridiculously badly underspecified and overspecified at the same time. |
| 19:41 | <Hixie> | it certainly is one of the more obvious choices |
| 19:42 | <Hixie> | more obvious than mathml, even |
| 19:42 | <jgraham__> | Oddly, I actually think it's the non-browser implementations of mathML that make it compelling |
| 19:43 | <Hixie> | here are formats with more non-browser implementations |
| 19:43 | <Hixie> | like, say, windows metafile |
| 19:44 | <jgraham__> | That's a vector language though |
| 19:44 | <Hixie> | oh sorry |
| 19:44 | <jgraham__> | SVG is implemented in browsers |
| 19:44 | <Hixie> | wrong one |
| 19:44 | <Hixie> | let me try again |
| 19:44 | <jgraham__> | MathML is implemented in editors |
| 19:44 | <Hixie> | "there are formats with more non-browser implementations" |
| 19:44 | <Hixie> | "like, say, pdf" |
| 19:44 | <Hixie> | or latex |
| 19:44 | <Hixie> | or whatever |
| 19:45 | <Hixie> | svg is implemented in some editors too |
| 19:45 | <Hixie> | i'm amused though that some people are so insecure in the inherent superiority of their technologies that they feel the need to do last minute canvasing |
| 19:45 | <jgraham__> | Well the advantage of MathML over PDF is that it roundtrips better and the advantage over LaTeX is that it is DOM compatible |
| 19:46 | <jgraham__> | (also with LaTeX you need to define a subset that you expect to work since it is extensible) |
| 19:46 | <Hixie> | there are pros and cons in both directions, yes |
| 19:47 | <Hixie> | as editor it's my role to consider them all |
| 19:47 | <Hixie> | :-) |
| 19:47 | <jgraham__> | Sure |
| 19:48 | <Philip`> | Is StarMath defined anywhere? |
| 19:48 | <Philip`> | (I assume that avoids the legacy LaTeX issues (like being based on an entire programming language)) |
| 20:27 | <Hixie> | well openoffice's mathml support is worthless |
| 20:27 | <Hixie> | it can't import mathml |
| 20:27 | <Hixie> | anyone have microsoft word? |
| 20:32 | <Philip`> | It can at least import MathML from inside ODF files |
| 20:32 | <mitsuhiko> | Hixie: the funniest thing about svg is that there still is no cross implementation way to express multiple whitespace beside nonbreaking spaces |
| 20:33 | <Hixie> | Philip`: no, it can't. it can ignore mathml, and import the staroffice format that's next to the mathml. |
| 20:33 | <mitsuhiko> | 1995 is calling and wants it's conventions back |
| 20:33 | <Hixie> | mitsuhiko: they don't support white-space? |
| 20:33 | <Philip`> | Hixie: If you delete the StarMath it can import the MathML |
| 20:34 | <mitsuhiko> | Hixie: inkscape uses xml:space="preserve", rsvg uses css with white-space: forgotthevalue |
| 20:34 | <mitsuhiko> | inkscape doesn't support css, firefox reads neither |
| 20:35 | <mitsuhiko> | can't remember which implementation broke with what, but we ended up using nonbreaking spaces |
| 20:36 | <Hixie> | Philip`: i just tried and that didn't seem to work |
| 20:36 | <mitsuhiko> | Hixie: you should have tested that in acid3 :D |
| 20:36 | <Hixie> | mitsuhiko: you should have suggested it when i asked for tests :-) |
| 20:36 | <Philip`> | Hixie: I tried it a while ago and it did work :-p |
| 20:37 | <Hixie> | oh |
| 20:37 | <Hixie> | i guess i should try again |
| 20:37 | <mitsuhiko> | touche |
| 20:37 | <Philip`> | (and it reconstructed the StarMath based on the MathML) |
| 20:37 | <mitsuhiko> | whatwg gave me back hope in standards. your's don't suck ;D |
| 20:38 | <Philip`> | Oddly, OO.o's Open dialog lists "MathML 1.01 (*.mml)", but I can't get it to do anything than read the text content out of the XML file and ignore the markup entirely |
| 20:39 | <Philip`> | Also I can't get "Import Formula" to do anything at all |
| 20:40 | <Philip`> | Ah |
| 20:40 | <Philip`> | Needs <?xml... |
| 20:40 | <Philip`> | Now it imports MathML, and renders it and gives a human-editable StarMath version |
| 20:43 | <Philip`> | (It's very helpfully and completely silent when it fails to load an ill-formed XML file, too) |
| 20:44 | <Hixie> | i couldn't get import formula to do anything with mathml that didn't have starmath in it |
| 20:44 | <Hixie> | but you're right |
| 20:44 | <Hixie> | modifying the odf worked |
| 20:44 | <jgraham> | Openoffice UI is generally pretty wful |
| 20:44 | <mitsuhiko> | muhaha. that's great xD |
| 20:45 | <Philip`> | So I think OO.o isn't exactly going to become my number one choice for equation editing |
| 20:46 | <Hixie> | no |
| 20:46 | <Hixie> | me either |
| 20:57 | <Philip`> | Hixie: I have access to Word 2007 over rdesktop, in case that might help |
| 20:57 | <Hixie> | yeah |
| 20:57 | <Hixie> | any information about what it's mathml support is like |
| 20:57 | <Hixie> | especially import |
| 20:57 | <Hixie> | would be great |
| 20:59 | <Philip`> | Any hints on where the import button is? |
| 21:00 | <Hixie> | no idea |
| 21:02 | <Hixie> | hey Philip`, you're good with turning things into graphs and stuff |
| 21:03 | <Hixie> | any chance you have a tool already set up that can turn the disaster that is the mathml dtd into a pretty picture showing the allowed nesting? |
| 21:08 | <Philip`> | Hixie: Word 2007's equation editor has 'Equation Options' -> 'Copy (MathML|Linear Format) to the clipboard as plain text', and when selecting the MathML option (the default is Linear Format) it does indeed copy equations into MathML on the clipboard |
| 21:09 | <Hixie> | cool |
| 21:09 | <Hixie> | i assume it can't import then? |
| 21:09 | <Philip`> | Haven't got that far yet :-p |
| 21:10 | <Hixie> | :-) |
| 21:10 | <Philip`> | If I copy MathML as plain text on the clipboard, then ctrl+V inside the equation editor, it gets converted into proper equations |
| 21:10 | Hixie | doesn't understand william's last e-mail to www-math |
| 21:10 | <Hixie> | nice |
| 21:10 | <Hixie> | very nice |
| 21:11 | <Philip`> | and even if I'm not in any kind of equation mode, then paste MathML, it gets converted into proper equations |
| 21:12 | <Hixie> | sweet |
| 21:12 | <Philip`> | (The kind of thing it puts onto the clipboard is <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"; xmlns:m="http://schemas.openxmlformats.org/officeDocument/2006/math"><mml:mi>A</mml:mi><mml:mo>=</mml:mo><mml:mi>π</mml:mi><mml:msup><mml:mrow><mml:mi>r</mml:mi></mml:mrow><mml:mrow><mml:mn>2</mml:mn></mml:mrow></mml:msup></mml:math>;) |
| 21:13 | <Hixie> | yeah, oo.o outputs prefixed stuff too |
| 21:13 | <Hixie> | still, that's a step in the right direction |
| 21:14 | <Philip`> | (It accepts anything that's well-formed XML with a <math> root in the right namespace) |
| 21:15 | <Philip`> | (so far as I can tell) |
| 21:15 | <Philip`> | (and not anything else) |
| 21:15 | <hsivonen> | Hixie: did you try zapping the StarMath stuff and the cache objects and then reopening the ODF? IIRC someone said it imports MathML in that case. |
| 21:15 | <hsivonen> | oh. already covered in log |
| 21:15 | <Hixie> | hehe |
| 21:17 | <Philip`> | I can't see any way to import or export equations as files directly |
| 21:17 | <Hixie> | i really don't understand http://lists.w3.org/Archives/Public/www-math/2008Mar/0011.html |
| 21:17 | <Hixie> | oh well |
| 21:17 | <Hixie> | Philip`: copy/paste is probably more than enough |
| 21:18 | <Hixie> | someone's gonna have to write a prefixed-xml to html convertor if we do this, though |
| 21:18 | hsivonen | might eventually do it as a side effect of doing something else... |
| 21:20 | <hsivonen> | Hixie: if you spec text/html serialization algorithm for MathML in text/html and SVG in text/html (in a reasonable way :-), I think I'm going to implement that as part of my SAX to HTML5 serializer |
| 21:21 | <hsivonen> | (well, I might implement it even without a spec. :-) |
| 21:22 | <hsivonen> | (reasonable way being streamable) |
| 21:22 | <Hixie> | it'll be streamable |
| 21:22 | <hsivonen> | cool |
| 21:22 | <Hixie> | the one thing i don't think we can do is switch to another insertion mode |
| 21:23 | <Hixie> | e.g. if we do: |
| 21:23 | <Hixie> | <svg><table></svg> -> <svg:svg><svg:table/></svg:svg> |
| 21:23 | <Hixie> | instead of: |
| 21:24 | <Hixie> | <svg><table></svg> -> <svg:svg></svg:svg><html:table/> |
| 21:24 | <Hixie> | or: |
| 21:24 | <Hixie> | <svg><table></svg> -> <svg:svg><html:table/></svg:svg> |
| 21:24 | <hsivonen> | Hixie: so are you really going to hard-code the names of MathML and SVG elements? |
| 21:24 | <Hixie> | ...i think we'll end up with problems where existing pages trigger into this insertion mode and break old pages |
| 21:25 | <Hixie> | hsivonen: either that, or hard code the html elements, or some combination thereof |
| 21:25 | <Hixie> | we might be able to get away with making unknown elements use whatever the parent element's namespace is |
| 21:26 | <hsivonen> | in the case of mathml, you could set aside unknown elements starting with 'm' as MathML |
| 21:26 | <Hixie> | that sounds dangerous |
| 21:26 | <Hixie> | what if we want to introduce a <mix> element or something |
| 21:26 | <Hixie> | or <media> or whatever |
| 21:26 | <Hixie> | now in older UAs it's in a different namespace |
| 21:27 | <hsivonen> | namespaces suck |
| 21:27 | <hsivonen> | but yeah |
| 21:28 | <Hixie> | thanks for the link to the relaxng schema, btw |
| 21:28 | <hsivonen> | np |
| 21:29 | <Hixie> | i think i'm gonna just go through the dtd and do it by hand though, the relaxng schema was as confusing to me :-) |
| 21:30 | <hsivonen> | there might be a RELAX NG visualizer in the latest oXygen. I'm not sure |
| 21:30 | hsivonen | finds RELAX NG Compact Syntax nicer than DTD |
| 21:31 | <Hixie> | the dtd syntax is horrible |
| 21:31 | <Hixie> | but it's the one in the spec :-) |
| 21:32 | <hsivonen> | fantasai's way of organizing RNC is much better than the way used in the MathML schema, though |
| 21:33 | <hsivonen> | that is, the MathML schema isn't as readable an the HTML5 schema |
| 21:34 | <hsivonen> | Hixie: fwiw, I think using annotation/annotation-xml for alternative serialization is bad |
| 21:34 | <hsivonen> | hijacking it for SVG seems workable, though |
| 21:34 | <Hixie> | i don't see the use case for having it at all |
| 21:34 | <Hixie> | we only want one presentation |
| 21:35 | <Philip`> | Hixie: I don't happen to have anything for parsing DTDs or converting them into anything interesting |
| 21:35 | <Hixie> | k |
| 21:35 | <Philip`> | (except for some ugly regexps that parse the HTML4 DTD, which seems like it wouldn't work in the MathML case) |
| 21:35 | <hsivonen> | Hixie: first example at http://golem.ph.utexas.edu/~distler/blog/archives/001475.html |
| 21:35 | <hsivonen> | Hixie: that's a Gecko-compatible DOM at least |
| 21:36 | <hsivonen> | Hixie: that one has only one representation |
| 21:36 | <Hixie> | hsivonen: why wouldn't we just have <math> <mtext> <svg>...</svg> </mtext> <mo> ... </mo> ... ? |
| 21:36 | <hsivonen> | Hixie: i.e. hijacked as an SVG escape hatch |
| 21:36 | <hsivonen> | Hixie: does that render in Gecko? |
| 21:36 | <Hixie> | none of this works in gecko, you can't put svg in text/html yet |
| 21:37 | <hsivonen> | Hixie: I want DOM consistency with XHTML+MathML+SVG |
| 21:37 | <Hixie> | sure |
| 21:37 | <Hixie> | i don't think requiring the UA to only do parsing changes is a requirement |
| 21:37 | <hsivonen> | Hixie: but yeah, the <semantics><annotation-xml> wrapper is silly, but that's a way that doesn't step on the toes of the MathML spec |
| 21:38 | <Hixie> | i'd rather just get teh spec changed |
| 21:39 | <Hixie> | <semantics> is more than just the escape hatch you see in that example |
| 21:39 | <Hixie> | it's also a way of including alternate formats, etc |
| 21:39 | hsivonen | tests |
| 21:39 | <Hixie> | and i don't think the semantics even really make sense there |
| 21:39 | <hsivonen> | the alternative format part seems harmful |
| 21:39 | <hsivonen> | witness OpenOffice.org |
| 21:39 | <Hixie> | yep |
| 21:40 | <Hixie> | anyone got something that does entity expansions in dtds? |
| 21:41 | <Hixie> | this dtd is insane |
| 21:42 | <hsivonen> | RNC looking better already? :-) |
| 21:43 | <Hixie> | yes |
| 21:43 | <Hixie> | i am in fact, as we speak, copying your subversion command line |
| 21:43 | <Hixie> | :-) |
| 21:43 | <Hixie> | (i had just looked at the site before) |
| 21:45 | <hsivonen> | Hixie: ok, <svg> in MathML works in Gecko without the <semantics><annotation-xml> cruft, so I agree with changing the spec |
| 21:45 | <Hixie> | heh |
| 21:45 | <Hixie> | ok so walk me through this schema |
| 21:46 | <Hixie> | what does this mean: |
| 21:46 | <Hixie> | mml.notprsubset.qname = element notprsubset { empty, mml.operators.common.attrib } |
| 21:46 | <hsivonen> | it declares a named production mml.notprsubset.qname |
| 21:46 | <hsivonen> | the convention is mml.elementname.qname in this particular schema |
| 21:47 | <hsivonen> | the production is an element named notprsubset |
| 21:47 | <hsivonen> | which has a content model of empty followed by the named production mml.operators.common.attrib |
| 21:47 | <hsivonen> | attributes are part of the content model in RELAX NG |
| 21:47 | <hsivonen> | so here the actual element-wise content model is empty |
| 21:48 | <hsivonen> | and the attributes are whatever mml.operators.common.attrib expands to |
| 21:48 | <Hixie> | aha |
| 21:48 | <Hixie> | ok |
| 21:48 | <hsivonen> | so element and empty are reserved words of RNC itself |
| 21:48 | <Hixie> | i need to strip all this attribute stuff |
| 21:49 | <hsivonen> | that *might* be easier to do with the XML syntax |
| 21:49 | <Hixie> | heh |
| 21:49 | <hsivonen> | but the XML syntax is seriously not human-readable |
| 21:49 | <Hixie> | this is not turning into a successful saturday :-P |
| 21:50 | <Hixie> | i couldn't even find the allowable content models in the spec other than in the dtd |
| 21:50 | <Hixie> | you'd think a spec would define that kind of stuff |
| 21:50 | <hsivonen> | no comment |
| 21:52 | <hsivonen> | Hixie: http://relaxng.org/compact-tutorial-20030326.html is pretty good and relatively short |
| 21:53 | <Hixie> | yeah i think i have the syntax figured out |
| 21:53 | <Hixie> | it's just a pain to process |
| 21:53 | <hsivonen> | You can convert between RNG and RNC with Trang |
| 21:53 | <hsivonen> | RNG can be processed with XML tools |
| 21:53 | <hsivonen> | yay for XML tools! |
| 21:54 | <Hixie> | uh huh |
| 21:54 | <Hixie> | i wish rnc had less ambiguous statment ends |
| 21:58 | jgraham | wonders what Dave Orchard is doing CCing public-html on that sort of mail |
| 22:02 | <hsivonen> | I thought WAF's use cases and reqs were just not written down |
| 22:18 | <Hixie> | wow, the mathml people are as insecure in the inherit superiority of their spec as the svg people |
| 22:19 | <hsivonen> | Hixie: how so? |
| 22:19 | <hsivonen> | Hixie: saying that existing implementations matter is not insecurity |
| 22:19 | <shepazu> | I can't speak for the mathml people, but your stated bias against SVG is what I have a problem with, not with the technical aspects of SVG versus other formats |
| 22:20 | <Hixie> | hsivonen: being scared that mathml isn't the "main" option |
| 22:20 | <sayrer> | it is true that the W3C process can create a sense of entitlement |
| 22:20 | <Hixie> | shepazu: my bias against svg is due to the deficiencies of the format, as are my biases against vml, wmf, etc |
| 22:20 | <Hixie> | shepazu: one has to consider all the pros and cons of all the options |
| 22:21 | <shepazu> | I haven't heard you say "I hate VML" and "wmf will never make it into HTML" |
| 22:21 | <Hixie> | shepazu: yes dear. |
| 22:22 | <Hixie> | (seriously, i'm not going to let personal biases affect how i evaluate the options, relax already.) |
| 22:22 | <shepazu> | ah, excellent point |
| 22:22 | <hsivonen> | Hixie: I can understand that in theory all options should be considered, but in this case, exploring them all and then settling for MathML and SVG seems like a distraction |
| 22:23 | <jgraham> | shepazu: it strictly irrelevant what Ian's biases are or whether they subconsciously affect him, as long as he comes to the right decision in the end |
| 22:23 | <Hixie> | hsivonen: i'm not making decisions of this magnitude without careful consideration of all the options, that would just be dumb |
| 22:23 | <jgraham> | The problem is determining the right decision ;) |
| 22:23 | <Hixie> | clearly the math people don't understand the concept of implying tags |
| 22:23 | Hixie | replies to them reexplaining what he meant |
| 22:24 | <jgraham> | Hixie: To be fair the point that implied start tags are confusing is reasonable |
| 22:24 | <Hixie> | yes |
| 22:24 | <Hixie> | but so is the point that mathml is so obscenely verbose as to be ridiculous |
| 22:25 | <jgraham> | Oh, I agree with that too |
| 22:25 | <hsivonen> | Hixie: David Carlisle does have a very good point that implied elements are no good if authors don't see the tree they are styling |
| 22:25 | jgraham | has hand authored MathML in the past, but only once |
| 22:25 | <Hixie> | hsivonen: i dunno, seems to work ok with <tbody> |
| 22:25 | <hsivonen> | umm. No |
| 22:25 | <hsivonen> | do you recall the thread about that with Rob Burns? |
| 22:26 | <Hixie> | i don't think one can draw any conclusions from threads with rob burns |
| 22:26 | <jgraham> | (the reason I've only done it once is obviously because it was so hard) |
| 22:26 | <Hixie> | at least, not any conclusions that apply to humanity at large |
| 22:28 | <shepazu> | jgraham: I don't see how you can derive your conclusion from what you said... how can a stated bias not be relevant? |
| 22:32 | <jgraham> | shepazu: Reread what I said ( it is true but a little (though not entirely) facetious) |
| 22:32 | <shepazu> | ok |
| 23:24 | <jgraham> | hsivonen: I guess annotations in mathml have a valid use case if they encode a superset of the information in the mathml e.g. if mathematica uses the annotation to encode information it needs to convert the formula into something it can manipulate |
| 23:24 | <jgraham> | (I don't know if this is actually the case or not) |
| 23:25 | <jgraham> | I agree that you could have problems keeping the content in sync though |
| 23:29 | <Hixie> | vendor-specific extensions to the language are not a use case i have any intention of addressing |
| 23:29 | <Hixie> | especially with explicit markup |
| 23:34 | <jgraham> | Hixie: Yeah, the fact that these things are (presumably) proprietary is bad. But given the reality of the computer algebra landscape it's not surprising the mathml people decided to address that use case |
| 23:47 | <jgraham> | Hixie: "If a MathML-input-conformant application receives input containing one or more elements with an illegal number or type of attributes or child schemata, it should nonetheless attempt to render all the input in an intelligible way, i.e. to render normally those parts of the input that were valid, and to render error messages (rendered as if enclosed in an merror element) in place of invalid expressions." |
| 23:48 | <Hixie> | is the "i.e." meant to be "e.g."? If not, what's with the "attempt" part? |
| 23:49 | <jgraham> | Hixie: Dunno. |
| 23:49 | <Hixie> | mozilla doesn't implement that |
| 23:49 | <Hixie> | e.g. data:text/xml,<math xmlns="http://www.w3.org/1998/Math/MathML">3</math>; |
| 23:49 | <jgraham> | I guess it's informative although it's generally hard to tell with the mathml spec |
| 23:50 | <Hixie> | data:text/xml,<math xmlns="http://www.w3.org/1998/Math/MathML"><mfrac>3 2</mfrac></math> works |
| 23:50 | <Hixie> | as in, shows an error |
| 23:50 | <Hixie> | data:text/xml,<math xmlns="http://www.w3.org/1998/Math/MathML"><mrow>2</mrow></math>; doesn't |
| 23:51 | <Hixie> | nor does data:text/xml,<math xmlns="http://www.w3.org/1998/Math/MathML"><mo><mo>+</mo></mo></math>; |
| 23:51 | <jgraham> | Section 7.2.1 seems to imply that anything to do with rendering is informative |
| 23:51 | <Hixie> | lovely |
| 23:52 | <jgraham> | Although roc had an example of something that is probably normative |
| 23:52 | <Hixie> | oh? |
| 23:52 | <jgraham> | the way that glyph size changes with script level |
| 23:52 | <Hixie> | ah yes |
| 23:52 | <jgraham> | and the CSS interactions |
| 23:53 | <Hixie> | you know, one could make a career out of rewriting these specs properly |
| 23:53 | <jgraham> | :) |
| 23:53 | <othermaciej> | Hixie: "could"? |
| 23:54 | jgraham | wonders if he should suggest that MathML 3 use RFC 2119 keywords so it's at least possible to tell what was supposed to be a conformance criterion |
| 23:54 | <Hixie> | othermaciej: i've only done one so far |
| 23:54 | <othermaciej> | jgraham: you should |