| 07:04 | <Ms2ger> | jgraham, yt? |
| 07:12 | <matjas> | zcorpan: wtf :'( |
| 07:14 | <matjas> | zcorpan: is XML allowed? |
| 07:17 | <zcorpan> | matjas: it's not #xmlpubquiz :-P |
| 07:17 | <matjas> | fffffuuuu |
| 07:17 | <zcorpan> | matjas: but i could go there i suppose... |
| 07:18 | <matjas> | data:text/xml,<x/> |
| 07:18 | <matjas> | document.documentElement.innerHTML = '<head/><head/><body/><body/>'; |
| 07:18 | <Ms2ger> | Hah |
| 07:19 | <matjas> | the Siamese DOM doesn’t show up in Dragonfly, but it does in the Chrome dev tools: http://i.imgur.com/xa9L5.png |
| 07:20 | <Ms2ger> | document.head.outerHTML = "." |
| 07:20 | <Ms2ger> | document.body.outerHTML = " " |
| 07:21 | <matjas> | Ms2ger: what the actual fuck |
| 07:22 | Ms2ger | bows |
| 07:23 | <Ms2ger> | zcorpan, do you know who I should bug about http://dvcs.w3.org/hg/html/file/0bc4307c200a/tests/submission/Opera/microdata/001.html ? |
| 07:26 | <zcorpan> | Ms2ger: you mean, who wrote the tests? |
| 07:26 | <Ms2ger> | Or who could review changes |
| 07:27 | <MikeSmith> | matjas: merged your pull request to add links to the TC39 HTML-formated spec |
| 07:27 | <zcorpan> | i've reviewed that test before so i guess i can do it again |
| 07:27 | <MikeSmith> | matjas: sorry for taking so long |
| 07:27 | <matjas> | MikeSmith: yay! thanks |
| 07:28 | <zcorpan> | Ms2ger: the last two checkins? |
| 07:28 | <Ms2ger> | zcorpan, no, a patch from David Zbarsky in our bug |
| 07:28 | <Ms2ger> | https://bugzilla.mozilla.org/show_bug.cgi?id=591467 |
| 07:31 | <zcorpan> | there's a David Zbarsky? did dbaron and bz merge? |
| 07:33 | <padenot> | bz's brother |
| 07:34 | <zcorpan> | i see |
| 07:34 | <jgraham> | Although the idea of dbaron and bz merging is quite terrifying |
| 07:35 | <zcorpan> | Ms2ger: is https://bugzilla.mozilla.org/attachment.cgi?id=627852&action=diff what i should be looking at? |
| 07:35 | <Ms2ger> | Yep |
| 07:36 | <Ms2ger> | jgraham, not the least for us, do you know how much work they get done? :) |
| 07:36 | <zcorpan> | line 6 and 7 is not OK |
| 07:36 | <Ms2ger> | Yeah |
| 07:37 | <jgraham> | Ms2ger: Maybe if they merged it would create someone who was 4 times as productive |
| 07:37 | <jgraham> | A super browser-engine god to rule over us all |
| 07:38 | <Ms2ger> | I think trying to rope in bz's two other brothers might be safer :) |
| 07:38 | <matjas> | zcorpan: the old line 6 and 7 weren’t okay as per http://www.w3.org/html/wg/wiki/Testing/Authoring/#Javascript_tests either (no type="text/javascript") |
| 07:39 | <matjas> | (all this is new to me) |
| 07:40 | <jgraham> | type is really not a requirement |
| 07:44 | <Ms2ger> | jgraham, the wiki page can be read as forbidding type |
| 07:44 | <jgraham> | matjas: (but I don't see what on that page you are referring to) |
| 07:44 | <jgraham> | Oh |
| 07:44 | <matjas> | jgraham: “You need the exact following code in your source:” |
| 07:45 | <matjas> | reading some other pages on the wiki, i get the idea they’re pretty strict about the test harness stuff |
| 07:45 | <jgraham> | That is a lie to children |
| 07:45 | <jgraham> | You need something that is semantically equivalent to that exact code |
| 07:45 | <gsnedders> | Well, thankfully we have no children contributing tests. |
| 07:45 | <jgraham> | I have better things to do in life than complain at people because they put useless legacy type attributes on their script elements |
| 07:48 | <jgraham> | gsnedders: The notion that lies to children only apply to children is itself a lie to children |
| 07:49 | <Ms2ger> | gsnedders, you don't write tests? ;) |
| 07:50 | <jgraham> | Turns out that gsnedders is a <del>criminal</del><ins>adult</ins> in the eyes of the law. |
| 07:51 | <jgraham> | Who knew? |
| 07:56 | <Ms2ger> | matjas, something like assert_equals(result, glyph, 'Expected ' + reference + ' to match ' ...) should work |
| 07:57 | <matjas> | Ms2ger: but e.g. in Trident you’d get "Expected ` ` to match ` `", which is not helpful at all as the glyphs look alike |
| 07:58 | <matjas> | Ms2ger: currently i print “Expected to match U+000A0 ( ); got U+00020 ( ) instead.” which is more useful |
| 07:58 | <Ms2ger> | Right |
| 07:59 | <zcorpan> | just do assert_equals(result, glyph, 'ALIENS') |
| 07:59 | <Ms2ger> | You should be able to pass "Expected to match U+000A0 ( ); got U+00020 ( ) instead." to assert_equals |
| 07:59 | <jgraham> | matjas: I am still somewhat confused as to how this is different from the html5lib test, aprt from (probably) being much faster to execute and easier to read |
| 08:00 | <matjas> | jgraham: it seems the html5lib test didn’t detect the Æ bug in WebKit, nor the 3 Trident bugs (?) |
| 08:01 | <jgraham> | That is quite worrying |
| 08:02 | <matjas> | not sure how that happened. my test is a simple 1:1 copy of this table http://www.whatwg.org/specs/web-apps/current-work/multipage/named-character-references.html |
| 08:02 | <jgraham> | Random feedback: please try to stick to one standardised property to get information; in general tests shouldn't have UA-specific codepaths |
| 08:02 | <matjas> | perhaps html5lib is missing a few? |
| 08:03 | <jgraham> | Even if that means that some browsers fail for reasons unrelated to the original test |
| 08:04 | <matjas> | jgraham: ok, I’ll keep that in mind when submitting. (in this case, it’s not really UA-specific, but rather feature-specific code (no UA sniffing)) |
| 08:05 | <jgraham> | Right, but it's still bad |
| 08:05 | <jgraham> | Ideally each browser should execute exactly the same code |
| 08:05 | matjas | “ideally” :) |
| 08:05 | <jgraham> | http://code.google.com/p/html5lib/source/browse/testdata/tokenizer/namedEntities.test is the source data for html5lib |
| 08:05 | <zcorpan> | Ms2ger: it's quite a big diff :-/ |
| 08:06 | <jgraham> | matjas: "ideally" as in "I will loudly complain if that isn't the case" :) |
| 08:07 | <gsnedders> | And when jgraham complains, he goes on and on and on. |
| 08:07 | <gsnedders> | You don't want him complaining at you. |
| 08:08 | <matjas> | gsnedders: noted :) |
| 08:09 | <matjas> | jgraham: aren’t the html5lib tests included in WebKit’s layout tests? I wonder how the Æ bug went unnoticed, not to mention the 93 other named char ref bugs that latest stable Safari has |
| 08:10 | <jgraham> | matjas: I am not sure what the WebKit people do. I thought they used the html5lib data but in a different harness |
| 08:11 | <annevk> | matjas: are those 93 by any chance references for two code points? |
| 08:11 | <matjas> | annevk: yeah |
| 08:11 | <annevk> | matjas: because that's a bug that was fixed fairly recently in WebKit |
| 08:12 | <matjas> | annevk: i know… but i wonder why they didn’t fix it sooner, given that the html5lib tests from 2010 show the errors |
| 08:12 | <matjas> | annevk: same for Æ which only got fixed the other day |
| 08:12 | <annevk> | matjas: because not all issues are considered of equal importance? |
| 08:12 | <annevk> | matjas: returning two code points prolly required an updated data structure |
| 08:13 | <matjas> | annevk: maybe, but no bug reports either? seems weird |
| 08:13 | <matjas> | (no bug reports for &AELig, that is) |
| 08:13 | <annevk> | matjas: oh, dunno about &AELig |
| 08:13 | <annevk> | abarth might know |
| 08:14 | <abarth> | hi |
| 08:14 | <gsnedders> | https://bugs.webkit.org/show_bug.cgi?id=87465 ? |
| 08:14 | <gsnedders> | Ah, your bug. |
| 08:14 | <abarth> | yeah, i screwed up my python falsy checks |
| 08:14 | <abarth> | people really like Python, but falsy is tricksy |
| 08:14 | <abarth> | anyway, it's fixed now :) |
| 08:15 | <abarth> | if you find these sorts of bugs, please feel free to fix them |
| 08:15 | <matjas> | abarth: we were wondering how this didn’t get detected by the layout tests sooner… aren’t the html5lib tests included in the WebKit layout tests? |
| 08:15 | <gsnedders> | The question was why the existing tests for entities didn't find it |
| 08:16 | <abarth> | i'm sure it was detected, but there are many, many bugs to fix |
| 08:16 | <abarth> | and not enough folks to fix them |
| 08:16 | <abarth> | if you'd like to help out by fixing bugs |
| 08:16 | <abarth> | i'm happy to find some that you might be interested in |
| 08:16 | <matjas> | abarth: i assumed that if a test from 2010 shows the error, _someone_ would have logged it in the bug tracker (even if it wouldn’t get fixed right away, which would be understandable) |
| 08:17 | <abarth> | oh, we have bugs on file from 2005 that haven't been fixed too |
| 08:17 | <abarth> | i tried fixing all the bugs in the order they were reported once |
| 08:17 | <matjas> | abarth: yeah, on file |
| 08:17 | <abarth> | that was fun for a week :) |
| 08:17 | <abarth> | I fixed some really crazy stuff |
| 08:17 | <matjas> | abarth: but this one wasn’t reported before, and i just wonder how that happened |
| 08:17 | <abarth> | but that turns out not to be the most productive order |
| 08:17 | <abarth> | i guess I don't understand the question you're asking |
| 08:18 | <abarth> | have you worked on a large software project before? |
| 08:19 | <gsnedders> | abarth: I think the point is that tests in the html5lib testsuite failed yet there was no bug report. i.e., WebKit failed its own tests without having any bug report about doing so. |
| 08:20 | <zcorpan> | Ms2ger: some tests are dropped. they should instead be changed to match what the spec expects (e.g. an exception for caller) |
| 08:20 | <abarth> | gsnedders: what's the point of filing a bug when there's a test that clearly shows the problem? |
| 08:21 | <abarth> | that's redundant, no? |
| 08:21 | <jgraham> | abarth: (depends on your testing setup and processes, really) |
| 08:22 | <abarth> | the truth is that this bug was vastly less important that many thousands of other bugs |
| 08:22 | <abarth> | so no one cared enough to file a bug about it or fix it until now |
| 08:22 | <zcorpan> | Ms2ger: after those two changes, i approve, but with reservation that we will defer more careful review until we update our impl to match the spec |
| 08:22 | <matjas> | abarth: ok, thanks! |
| 08:23 | <abarth> | if there are other bugs that you'd like to see fixed, please feel encouraged to file bugs about them and/or write patches to fix them |
| 08:25 | <zcorpan> | matjas: you could output your own log that you're using now as well as using the standard testharness output |
| 08:26 | <matjas> | zcorpan: i see, thanks! |
| 08:28 | <zcorpan> | though it would be nice to be able to implement your own custom message in the output table, maybe by using a function as the message argument |
| 08:49 | <smaug____> | huomenta |
| 09:29 | <AryehGregor> | matjas, I've checked in my editing spec tests to Gecko, so they're run as regression tests. If you think that means anyone has sat down and classified every single expected failure, you're crazy. :) |
| 09:29 | <AryehGregor> | Likewise richtext2 has been part of the Gecko test suite for some time. |
| 09:30 | <AryehGregor> | But there are tons of expected failures, no one has looked at all of them or filed bugs on all of them. |
| 09:30 | <AryehGregor> | It's not clear in many cases why they're failing or whether the test suite is even correct. |
| 09:30 | <matjas> | AryehGregor: not necessarily a separate bug for every single failure, but I would expect a “master tracking bug” of some sort, yeah |
| 09:30 | <matjas> | I guess I’m just naive. |
| 09:31 | <AryehGregor> | Or maybe you're volunteering to review all expected failures in Gecko/WebKit's test suites and file bugs on all of them? :) |
| 09:31 | <AryehGregor> | Generally you can't even file a useful bug unless you know the spec and/or code well and spend a bunch of time debugging -- the failure is often nontrivial. |
| 09:33 | <AryehGregor> | There are tons of low-hanging editing test failures that I haven't looked at because I have lots of other stuff to do. |
| 09:33 | <AryehGregor> | And in my case, I'm even the one who's paid by Mozilla both to maintain the spec/tests and to write editing code for them. |
| 09:34 | <AryehGregor> | Although I wrote the tests while I was working for Google -- my CSS Transforms tests, which I wrote while at Mozilla, have bugs filed for all Gecko failures. |
| 09:34 | <AryehGregor> | (a number of which are still open) |
| 09:39 | <Ms2ger> | zcorpan, thanks! |
| 10:03 | <jgraham> | matjas: Yeah, it would be a huge amount of work to make a bug report for every single failing test, particularly from imported test suites which can be of very dubious quality |
| 10:03 | <zcorpan> | Ms2ger: welcome. in return, maybe you could convert the web worker testsuite to use testharness? :-P |
| 10:04 | <MikeSmith> | does anybody know what the term "browse-by Web" is supposed to mean? |
| 10:04 | <matjas> | gsnedders: did Steven Levithan end up filing a bug detailing the non-standard regex additions in Opera? if not, here’s a list: http://kangax.github.com/es5-compat-table/non-standard/ |
| 10:07 | <annevk> | matjas: maybe link that from http://wiki.whatwg.org/wiki/Web_ECMAScript#RegExp ? |
| 10:07 | <zcorpan> | there's already a link |
| 10:08 | <annevk> | oh right |
| 10:08 | <matjas> | ah, at the bottom, damn |
| 10:08 | <matjas> | too late |
| 10:08 | <Philip`> | MikeSmith: Maybe the parts of the web where you only stop at a page for a short time and glance over it and then browse on by, unlike e.g. Gmail or Facebook which are more like major destinations you'll spend hours at |
| 10:09 | <MikeSmith> | Philip`: yeah, that would seem like an intuitive definition, given those words |
| 10:09 | <gsnedders> | matjas: He did. |
| 10:10 | <annevk> | matjas: reverted |
| 10:10 | <MikeSmith> | Philip`: but looking at http://www.w3.org/2012/05/sysapps-wg-charter.html I don't see what connection "browse-by Web" has to the scope of that proposed work |
| 10:10 | <gsnedders> | matjas: And that's nothing compared with what we support |
| 10:10 | <annevk> | matjas: also sorry :) |
| 10:10 | <jgraham> | MikeSmith: Maybe it's the bits of the web you would like to shoot? |
| 10:11 | <MikeSmith> | heh |
| 10:11 | <matjas> | gsnedders: now i’m curious |
| 10:11 | <MikeSmith> | jgraham: that would be a lot of bits |
| 10:11 | <gsnedders> | matjas: Named capturing groups, nested character classes, etc. |
| 10:13 | <Philip`> | MikeSmith: Yeah, in that case it sounds like they're saying "the "browse-by" web" for what everyone calls "the web", but they don't want to admit that their adoption of web-originated technologies for use as the core platform of new devices with very different characteristics is not really the web |
| 10:14 | <MikeSmith> | yeah |
| 10:14 | <MikeSmith> | exactly |
| 10:15 | <MikeSmith> | fwiw, I suggested using "Web-application-unsafe APIs" instead |
| 10:15 | <MikeSmith> | http://lists.w3.org/Archives/Public/public-sysapps/2012May/0028.html |
| 10:15 | <MikeSmith> | or "non-SOP APIs" |
| 10:16 | <annevk> | MikeSmith: did you see my comment earlier about https://www.w3.org/Bugs/Public/show_bug.cgi?id=17205 being actual spam? |
| 10:16 | <MikeSmith> | annevk: yeah |
| 10:16 | <Philip`> | I suppose the problem is that "web" has become a generic brand name for a collection of technologies, rather than being about a distributed network of interlinked resources or whatever the original idea was |
| 10:16 | <annevk> | MikeSmith: though it seems kind of weird spam looking at it closer; URL returns some JSON |
| 10:17 | <Philip`> | and those technologies happen to be just about good enough that people want to reuse them in totally different contexts |
| 10:17 | <annevk> | MikeSmith: fine if I close that bug? |
| 10:17 | <MikeSmith> | annevk: you want me to mark that account as a spammer? Looking at that bug, it's not clear to me that it's intentional malicious spamming |
| 10:17 | <MikeSmith> | annevk: yeah, please close it |
| 10:18 | <annevk> | MikeSmith: can only find that bug for that user |
| 10:18 | <annevk> | but yeah, lets wait a bit |
| 10:20 | <MikeSmith> | OK |
| 10:22 | <Philip`> | MikeSmith: I think "non-SOP APIs" is good, because then the rest of the web can be called "soppy APIs" |
| 10:25 | <gsnedders> | https://developer.mozilla.org/en/JavaScript/Reference/Global_Objects/WeakMap — I wonder how much effect that warning will really have… |
| 10:26 | <zcorpan> | non-SOAP APIs? |
| 10:33 | <Ms2ger> | zcorpan, :) |
| 10:34 | <Ms2ger> | zcorpan, might be easier if you threw jsframework.js my way ;) |
| 10:35 | <zcorpan> | that's probably doable |
| 10:36 | <Ms2ger> | Also |
| 10:36 | <Ms2ger> | odinho, about that assert_throws patch... |
| 10:36 | <Ms2ger> | jgraham, about that timout patch... |
| 10:38 | <jgraham> | Ms2ger: Yeah, I should push that |
| 10:38 | <jgraham> | You had some feedback that I've forgotten |
| 10:39 | <Ms2ger> | Philip`, about that 2d.pattern.repeat.null bug... |
| 10:39 | <odinho> | Ms2ger: wawawa, I thought about that when I came to work today actually :P |
| 10:39 | <Ms2ger> | jgraham, You're inconsistent about if () { vs if ()\n{ |
| 10:40 | <Ms2ger> | Dunno what else I'd complain about |
| 10:40 | <Ms2ger> | odinho, I'm glad to hear that :) |
| 10:42 | <zcorpan> | Ms2ger: http://simon.html5.org/dump/jsframework.js |
| 10:43 | <Ms2ger> | Thanks, I'll have a look at some point |
| 10:44 | <zcorpan> | Ms2ger: awesome, thanks! |
| 10:44 | <odinho> | IDLharness is cool although I don't really trust it all that much yet. It's just so ... automatic and stuff. :P |
| 10:45 | <zcorpan> | Ms2ger: one thing i recall is that the web worker testsuite at places has nested tests, which is not considered a good practice. instead it should create all tests up front (as async tests). that way, the number of tests don't change based on the result of the tests |
| 10:45 | <Ms2ger> | Mm |
| 10:46 | <Ms2ger> | matjas, `\0` (U+0001) < eh? |
| 10:46 | <matjas> | Ms2ger: what? |
| 10:46 | AryehGregor | wants nsINode::AsContent |
| 10:47 | <Ms2ger> | AryehGregor, wrong window, and sure, file a bug :) |
| 10:47 | <AryehGregor> | Drat. |
| 10:47 | AryehGregor | usually has #developers here, gets mixed up when it's #whatwg |
| 10:47 | <Ms2ger> | matjas, surely that's U+0000 |
| 10:47 | <matjas> | Ms2ger: yeah; where did i make that mistake? |
| 10:47 | <Ms2ger> | The twattersphere |
| 10:48 | <matjas> | whoops, thanks! |
| 10:50 | <AryehGregor> | odinho, I don't trust it so much either, and I wrote it. :) |
| 10:50 | <AryehGregor> | But it's cool! |
| 10:51 | <matjas> | zcorpan: on http://simon.html5.org/htmlpubquiz, could you please link to the explanation as well? |
| 10:52 | <zcorpan> | matjas: it shows up if you click on correct answer |
| 10:52 | <matjas> | zcorpan: not on a mobile device |
| 10:52 | <matjas> | (do you really want to rely on the current twitter web layout?) |
| 10:52 | <zcorpan> | blame twitter |
| 11:07 | <annevk> | zcorpan: relying on twitter to keep those relations indefinitely seems unwise though |
| 11:07 | <annevk> | zcorpan: anecdotal evidence suggests replies, retweets, etc. get lost over time |
| 11:08 | <zcorpan> | ok. i guess i should just move over all the data |
| 11:11 | <odinho> | Hmz, sicking did send an email not too long ago. But no IRC. |
| 11:11 | <odinho> | So, there's a double ;; in one of the IDL-fields in the idb spec, -- should I really make a bug for that? Feels like a real waste of bytes. |
| 11:13 | <annevk> | bugs are filed for typos all the time |
| 11:13 | <annevk> | the real waste is spam |
| 11:14 | <Ms2ger> | <meta name='revision' content='$Id: Overview.html,v 1.41 2010/03/24 17:58:17 nmehta3 Exp $'/> |
| 11:32 | <Stevef> | Philip: any data on mathml on the web? |
| 11:39 | <david_carlisle> | Stevef: what kind of data are you looking for? (just passing through and the last comment in the channel is on mathml:-) |
| 11:59 | <Stevef> | david_carlisle: looking into providing accessible math, wondering how much mathml is used and if prose form of math is provided alongside for AT users |
| 11:59 | <annevk> | shouldn't AT support math? |
| 12:00 | <annevk> | seems better than require everyone to do make work |
| 12:01 | <Stevef> | annevk: not that simple even if it is supported still a problem for vision impaired to understand complex equations, also browsers need to support math, support is spotty |
| 12:02 | <jgraham> | No one is going to type out a verbal description of all their formulae |
| 12:02 | <jgraham> | Since there is support for TeX->speech it seems like presentation mathml to speech should be just as possible |
| 12:03 | <jgraham> | Anyway, what I actually wanted to say is "does anyone know anything about navigation"? |
| 12:04 | <Stevef> | jgraham: for educators it is an issue, there is extensive research on the topic, simply saying "no one will" is not helpful |
| 12:04 | <jgraham> | It seems like Chrome and Firefox abort navigations triggered from inside the "prompt to unload" algorithm i.e. in the beforeunload event |
| 12:05 | <Stevef> | there is editing software that outputs MathSpeak i.e. equations in prose form |
| 12:05 | <jgraham> | But I can't see any requirement for this in the spec. Anyone know if there's a reason for the difference or if it is just a bug? |
| 12:06 | <jgraham> | Stevef: Using software to automate the translation is equivalent to having it built in to the browser. I think working from the assumption that no one will write out verbal descriptions is an excellent position because it forces you to consider solutions that will work with the vast majority of content rather than just solutions where people put in special additional effort |
| 12:06 | <david_carlisle> | Stevef: There is a lot of work being done at present in DAISY and NIMAS groups on refining the guidelines on how MathML should interact with AT software, If you ask on www-math list rather than here you should be able to hook up with other people working on that |
| 12:07 | <Stevef> | david_carlisle: thanks |
| 12:12 | <Stevef> | jgarham: and thats what I am trying to undesrstand what are the mechanisms for providing accessible math and are prose transaltions provided (from automated or other source) |
| 12:13 | <Stevef> | jgraham: because MathML is not a complete solution |
| 12:16 | <Stevef> | jgraham: saying 'nobody' will is incorrect, saying the vast majority won't is plausible |
| 12:18 | <david_carlisle> | Stevef: Have you tried MathPlayers AT cpabilities? That's probably the most generally available system in that area, the version in the browser has an option to read the expression, but I understand that that is really just a demo on how it could hook into a general AT framework (but it isn't really my area) |
| 12:18 | <jgraham> | I agree that "nobody" isn't strictly accurate. But acting as if it is seems more likely to produce good results than optimising for the difference |
| 12:21 | <Stevef> | david_carlisle: have been playing with MathPlayer, the built in speech capability is good, supports many languages. limitation is that it only runs in IE, JAWS/Window YES/zoomtext reader/magnifier works well with it. |
| 12:22 | <Stevef> | jgraham: i am not advocating that all math must be accompanied by hand written prose form of equation, I am trying to work out how people do or don't do math on the web |
| 12:23 | <Stevef> | jgraham: in particular accessible math |
| 12:25 | <david_carlisle> | Stevef: Yes I know it's windows only, but having _an_ implementation is good to validate the general premise that it is _possible_ to get a good accessible behaviour from the markup. After that it's "just" a matter of applying peer pressure (or citing relevant legislation:-) to encourage others to follow suit. |
| 12:26 | <jgraham> | I don't really know how accessible maths is different from any other kind. I mean I imagine if I was a blind physicist, for example, I would want/need to have access to all the same maths content as my peers |
| 12:27 | <Stevef> | david_carlisle:true, something is better than nothing, just trying to work out the practicalities for a online education company who need to provide accessible math |
| 12:28 | <Stevef> | jgraham:point being? also I am not talking about blind physicist I am talking about providing content to a school kid |
| 12:28 | <david_carlisle> | jgraham: True enough but the mathml spec does try to give general guidance on how to make the markup more accessible including things that have no effect on the visual presentation, things like the invisible unicode characters fro invisible times or function application |
| 12:32 | <david_carlisle> | Stevef: we have members of the math WG who work full time on these aspects, but I'm not one of them so I can only pass on second hand information I've picked up over the years, which is why I suggested www-math might be a better forum if you want to get first-hand reports of what people are doing |
| 12:32 | <Stevef> | david_carlisle: sure, just asked philip as i know he collects data on stuff |
| 12:34 | <david_carlisle> | Stevef: Yes if you want to reach people with real data collected from real web sites this probably is the place:-) |
| 12:37 | <Stevef> | david_carlisle: also seems that what is published is presentational mathML rather than content mathml, so meaning is not fully convyed in the markup exposed to AT/ |
| 12:40 | <david_carlisle> | Stevef: mostly perhaps yes, by coincidence I got pointed to this message this morning which is relevant: the thread starts here, but includes a quote from Nemeth argguing presentation is what you want: http://host.nfbnet.org/pipermail/blindmath_nfbnet.org/2012-May/005435.html |
| 12:42 | <Stevef> | david_carlisle: cheers, the subject of accessible math is a real can of worms, it is difficult to know what advice to give to content producers |
| 12:43 | <jgraham> | annevk, Ms2ger: DOM should define the term "empty text node" |
| 12:44 | <david_carlisle> | Stevef: It's probably not fair to say meaning isn't conveyed to AT as that implies that meaning is being lost. If the same markup is being used for visual rendering and for AT then that isn't the case. It will always be the case that there is more presentation mathml (even if browsers supported content mathml natively) as it is much easier to produce, you can give a hand drawn expression to... |
| 12:44 | <david_carlisle> | ...a suitably trained typist and get presentation mathml, but to produce content mathml you need to know something about what it means., which means production has to be a lot more controlled |
| 12:45 | <Ms2ger> | jgraham, file a bug, please? |
| 12:45 | <Stevef> | david_carlisle:ok thanks for the clarification |
| 12:46 | <annevk> | jgraham: also list a use case in that bug, please |
| 12:47 | <jgraham> | annevk: For the definition? HTML uses it |
| 12:47 | <Ms2ger> | HTML also defines it, fwiw |
| 12:49 | <jgraham> | Where? |
| 12:50 | <Ms2ger> | At the end of 2.1.3 DOM trees |
| 12:50 | <jgraham> | Interesting |
| 12:51 | <jgraham> | That should really be in DOM I think |
| 12:51 | <jgraham> | But more importantly Hixie should link to the definition |
| 12:53 | <annevk> | I'm happy to have the collection of useful terms related to the DOM in DOM |
| 12:53 | <annevk> | that makes a lot of sense |
| 12:53 | <annevk> | I similarly asked dglazkov to file a bug on getting DOM to define subtree |
| 14:40 | <Stevef> | jgraham: Tex typesetter and text translation http://www.math.union.edu/~dpvc/transfer/mathjax/speech-lab.html |
| 14:48 | <jgraham> | Stevef: Seems pretty cool |
| 14:48 | <david_carlisle> | Stevef: given the URI that's presumably mathjax based so would work for MathML too, as MathJax has two input parsers, one for TeX-like syntax and one for MathML. |
| 14:50 | <Stevef> | right, got the link from this guy https://twitter.com/#!/pkrautz |
| 14:53 | <david_carlisle> | Stevef: Davide has a description of it here: http://lists.w3.org/Archives/Public/public-fx/2012AprJun/0126.html |
| 15:06 | <Stevef> | david_carlisle:cheers interesting thread |
| 15:35 | <annevk> | roc: fwiw, http://www.w3.org/TR/from-origin/ is now updated |
| 15:35 | <annevk> | Ms2ger: are you going to review DOM 3 Events? should I? |
| 15:36 | <Ms2ger> | I'm planning to |
| 15:36 | <Ms2ger> | But the more the better :) |
| 15:37 | <annevk> | "HTTP living standard" would sure be welcome |
| 15:37 | <annevk> | seven part spec with god knows how many registries and extensions sucks monkey balls |
| 15:37 | <Ms2ger> | "IETF" |
| 16:16 | <dglazkov> | good morning, Whatwg! |
| 16:16 | <dglazkov> | annevk: please steal the file a bug button script from me for your spec, so that I can file bugs easily :) |
| 16:18 | <Ms2ger> | dglazkov, you mean the "file a bug" link he has at the top of all his specs? :) |
| 16:20 | <odinho> | Ms2ger: Prolly the just point and write and press enter. |
| 16:31 | <davidb> | at |
| 16:31 | <Ms2ger> | dot |
| 16:31 | <odinho> | bash: at: command not found |
| 16:32 | <Ms2ger> | Eh, let's discuss top-posting |
| 16:33 | <odinho> | ... crickets ... |
| 16:33 | <odinho> | and that's that. |
| 16:47 | davidb | chuckles |
| 17:15 | <jgraham> | irony alert: ifette complained about *other people's* mail formatting |
| 17:15 | <jgraham> | His is generally pretty terrible |
| 17:15 | <jgraham> | Not the worst I have ever seen but often hard/impossible to follow in alpine |
| 17:16 | <Ms2ger> | Also fun: nonstandard quoting characters |
| 17:16 | <Ms2ger> | (Hi, Alex Mogilewski) |
| 17:17 | <jgraham> | The worst offenders are 1) people who use lotus notes (i.e. IBM employees) and 2) people that top post (generally gmail users working at Google) |
| 17:17 | <TabAtkins> | Heh, yeah, I don't know what's that about. |
| 17:17 | <jgraham> | Seems like the bigger your company is the worse your mail habits are likely to be |
| 17:18 | <zewt> | the worst are people who post in light blue in a giant font |
| 17:18 | <jgraham> | (pretty sure it is notes that does the quotes-distinguished-by-colour thing) |
| 17:18 | <zewt> | people posting in large fonts always seem to me like they think their posts are more important than everyone else's |
| 17:19 | <zewt> | heh |
| 17:19 | <jgraham> | Well alpine doesn't do light blue or coloured fonts and I think I turned that off in thunderbird too |
| 17:19 | <jgraham> | So generally when people do that I have to disregard their message due to it being unintelligable |
| 17:19 | <zewt> | the colored-quoting thing is definitely worse, but fortunately fairly rare |
| 17:19 | <zewt> | at least on public lists |
| 17:20 | <TabAtkins> | That seems like a legitimate effect, actually. In a big company, most of your email will be internal. Most people aren't picky about their email, so they'll use whatever is the default and easiest. Thus, in a large company you're exposed much more to "bad" practices caused by monoculture. |
| 17:20 | <jgraham> | Sometimes I have had to unpick threads and only been able to work out who was saying what by looking for the place where the opinion inverted |
| 17:20 | <zewt> | also messages where people inline quote, but leave ten pages of quotes above *and* below their reply |
| 17:20 | <TabAtkins> | Whereas someone in a smaller company has to deal more with external emails in a variety of formats, so they get pickier and politer intheir formatting. |
| 17:20 | <TabAtkins> | zewt: Heh, see: every single internal thread at Google. |
| 17:21 | <TabAtkins> | It makes me die inside. |
| 17:21 | <zewt> | gmail will collapse both, but then replying is a complete mess |
| 17:21 | <jgraham> | Yeah, it seems that gmail encourages terrible habits |
| 17:21 | <zewt> | to be honest, i don't think it's worth trying to get non-technical users to inline quote if they don't do it intuitively, because they tend to make a hash of it |
| 17:22 | <jgraham> | Pretty sad coming from a company so closely associated with the internet |
| 17:22 | <TabAtkins> | zewt: I'd expect everyone on one of our lists to be technical, though. |
| 17:22 | <zewt> | sure |
| 17:23 | <hober> | All of this is made that much worse by having Ahem installed |
| 17:23 | <TabAtkins> | s/worse/better/ |
| 17:23 | <TabAtkins> | Because then you get to do fun ascii art. |
| 17:23 | <hober> | TabAtkins: :) |
| 17:23 | <zewt> | another side benefit of html mail: you can explicitly declare when text is meant to be fixed-width |
| 17:24 | <zewt> | rather than assuming everyone reads mail in fixed width (which most people don't) |
| 17:24 | <TabAtkins> | That, and inline images, are the only two reasons I ever switch to sending HTML email. |
| 17:24 | <jgraham> | hober: Why? Unless you only have ahem installed |
| 17:24 | <zewt> | italics and bold are pretty useful when you're discussing spec text |
| 17:25 | <hober> | jgraham: email from outlook read in mail.app on a machine with ahem installed is, shall we say, square. |
| 17:25 | <zewt> | reading specs without keywords marked is a headache |
| 17:25 | <TabAtkins> | Oh, so now it's not *cool* enough for you, hober? |
| 17:25 | <TabAtkins> | Jeez. |
| 17:25 | <hober> | jgraham: outlook's html email contains an invalid font-family rule which gets matched with ahem in webkit |
| 17:25 | <jgraham> | hober: Wow. Mind you, I like to imagine that Björn Höhrmann only has ahem installed |
| 17:26 | <jgraham> | and uses it for all things |
| 17:26 | <hober> | hahahahahahaa, that would explain a few things |
| 17:26 | <TabAtkins> | Hahahahaha |
| 17:26 | <TabAtkins> | Unrelated: how do I politely suggest to someone that they don't understand any of the details of what they're talking about? |
| 17:27 | <Ms2ger> | You don't, on the interwebs |
| 17:27 | <TabAtkins> | Specifically, John Daggett, in his discussion of Variables, keeps making assertions about CSS grammar that are simply wrong. |
| 17:27 | <Ms2ger> | Well, he thinks the same about your assertions :) |
| 17:27 | <TabAtkins> | I understand that not everyone can read FLEX or whatever easily, but still. |
| 17:28 | <jgraham> | TabAtkins: Make an implementation of the grammer that you can feed his mistakes to and demonstrate that they don't match? |
| 17:28 | <jgraham> | *grammar |
| 17:29 | <TabAtkins> | I was planning on doing that anyway, but I don't particularly feel like doing so *right now*. ^_^ |
| 17:29 | <TabAtkins> | Plus I'd do the implementation according to my Syntax draft, so I've misread the FLEX grammar, I'll still get it wrong. |
| 17:30 | <jgraham> | Yeah, but you could ask him to point out the bug :) |
| 17:31 | <TabAtkins> | Well, I'm pointing out his mistake anyway, so that's equivalent. |
| 17:32 | <jgraham> | It wasn't an entirely serious suggestion. If you can't do things in the normal way (i.e. say "this doesn't match the spec; as we se in section X.Y {explaination of the mismatch}" then I don't know what will work |
| 17:32 | <TabAtkins> | That's working individually, he's just made several incorrect assertions that are plainly wrong if you have a good understanding of the grammar. |
| 17:33 | <TabAtkins> | Like assuming that "counter\(foo\)" parses as a function (it's an ident). |
| 17:33 | <jgraham> | Ah, so the problem is how to say "I think you need to reread the grammar section of the spec" |
| 17:33 | <TabAtkins> | Yes. |
| 17:33 | <TabAtkins> | However, I said almost exactly that in the email I just sent, so whatever. |
| 17:49 | <zewt> | is there any difference visible to scripts between {} and {a: undefined}? |
| 17:50 | <zewt> | guess hasOwnProperty("a") returns true |
| 17:51 | <jgraham> | Yeah |
| 17:52 | <TabAtkins> | Yup, that's it. |
| 17:53 | <TabAtkins> | Pair that with a prototype walk just in case the property is further up the chain, and you can robustly detect it. |
| 17:53 | <zewt> | undefined is ... defined D: |
| 17:53 | <jgraham> | Well "a" in foo is just that |
| 17:53 | <TabAtkins> | However, good code shouldn't treat the two differently. |
| 17:54 | <TabAtkins> | If you really need a "this property exists but doesn't have a value", use null. |
| 17:54 | <TabAtkins> | That's what it's there for - to be the same as undefined, but detectably different. |
| 17:54 | <zewt> | well, yeah |
| 17:54 | <jgraham> | TabAtkins: That's like saying you should never use hasOwnProperty withotu also checking the value isn't undefined |
| 17:54 | <zewt> | (the question comes from something unrelated to web stuff) |
| 17:54 | <jgraham> | Which is crazy |
| 17:55 | <TabAtkins> | jgraham: You just shouldn't use hasOwnProperty. ^_^ for-in is a broken construct that encourages bad habits. |
| 17:55 | <jgraham> | For for...in |
| 17:55 | <jgraham> | s/For/Or/ |
| 17:55 | <TabAtkins> | Which is why we did it all again with for-of, which acts the way you'd *expect*, rather than the crazy way that for-in does. |
| 17:55 | <jgraham> | Or... I forget hwo array iterators work with sparse arrays |
| 17:56 | <TabAtkins> | Depends on the details of the iterator. Which one are you talking about? |
| 17:58 | <Ms2ger> | new Iterator(), duh |
| 17:58 | <TabAtkins> | Ah, thanks, that's useful. ^_^ |
| 19:06 | <jgraham> | Sigh. ifette claims that the fact that his mail appears broken in the archives and in other mail clients means that everyone else should switch clients/archiving software |
| 19:07 | <Ms2ger> | Duh? |
| 19:16 | <annevk> | jgraham: Gmail should really fix its text/plain handling imo |
| 19:17 | <annevk> | jgraham: I noticed the problem too; Opera's pretty awesome at handling quotes in emails (and reformatting them) |
| 19:25 | <Hixie> | i am back |
| 19:25 | <Hixie> | and have about 1000 e-mails to read |
| 19:25 | <Hixie> | anything urgent for me to do first? |
| 19:25 | <zewt> | the bathroom's that way --> |
| 19:33 | <annevk> | Hixie: nothing has collapsed so far :) |
| 19:39 | <Hixie> | cool |
| 19:41 | <Ms2ger> | The HTMLWG has requested that you add a link to a 404 error to the spec, if you were gone already when they did that |
| 19:46 | <Hixie> | come again? |
| 19:47 | <Hixie> | (264 e-mails on responsive images? jesus) |
| 19:49 | <Wilto> | Hixie: Florian Rivoal has a pretty great proposed compromise at the end. Y’know, after the first 263 chapters. |
| 19:58 | <Hixie> | Wilto: i will be reading all the e-mails as always :-) |
| 19:59 | <Wilto> | Hixie: Oh, definitely didn’t mean it as in “skip the rest.” Just throwin’ that out there. |
| 20:03 | <Hixie> | Ms2ger: i don't see any decisions in recent times |
| 20:04 | <Ms2ger> | https://www.w3.org/Bugs/Public/show_bug.cgi?id=11204#c34 |
| 20:08 | <Hixie> | lol |
| 20:09 | <jgraham> | Hixie: Nothing urgent, but the document loading / navigation parts of the spec turn out to have quite a few differences from implementations, and we would appreciate it if you could look at those relatively soon |
| 20:10 | <Hixie> | jgraham: is there mail? |
| 20:10 | <jgraham> | Hixie: Bugs mostly |
| 20:10 | <Hixie> | jgraham: i mean, i know the browsers all differ from each other, that's nothing new... |
| 20:10 | <jgraham> | Although there are some older mail threads |
| 20:10 | <Hixie> | can you send me a mail with links to all the things you want me to process on the topic? |
| 20:11 | <jgraham> | Hixie: There seem to be places where the spec is different from any browser and the browsers are relatively consistent |
| 20:11 | <Hixie> | well that seems bad |
| 20:11 | <jgraham> | Yeah, I will put together a mail tomorrow |
| 20:11 | <jgraham> | There is probably more to discover of course |
| 20:11 | <Hixie> | cool, thanks |
| 20:53 | <WeirdAl> | Hixie - I love the big black box on the XBL2 note :) |
| 20:54 | <WeirdAl> | voice of God, with Broadway lights |
| 20:54 | <Hixie> | url? |
| 20:55 | <WeirdAl> | http://www.w3.org/TR/2012/NOTE-xbl-20120524/ |
| 20:56 | <Hixie> | ah, that was below the fold when i brought it up, heh |
| 20:56 | <Hixie> | so i didn't see it |
| 20:59 | <WeirdAl> | it's a nice touch |
| 21:12 | <Hixie> | if anyone is interested in participating in this year's http://www.jseverywhere.org/ please let me know |
| 21:13 | <Hixie> | in other news, i do not understand how so many people manage to get an account and subscribe to parts of the HTML spec and yet not understand why they get e-mails when i change the spec |
| 22:23 | <Hixie> | abarth: i have some e-mails from you regarding https://bugs.webkit.org/show_bug.cgi?id=44641 -- should i just discard them? as you say in the bug, nobody else really got particularly excited about it |
| 22:28 | <zewt> | fwiw, not really sure how it's better than just <?= htmlspecialchars($untrusted_string) ?>, it's just more opaque and not human readable |
| 23:23 | <abarth> | Hixie: yeah, that's fine |
| 23:24 | <Hixie> | k, thanks |
