| 01:47 | <zewt> | i hate when conversations become two-headed simultaneous "should we do this" and "how might we do this", heh |
| 01:52 | <crocket> | hi |
| 01:53 | <crocket> | Is there a website for testing web browser's conformance to HTML 4? |
| 02:01 | <Philip`> | crocket: HTML4 was never defined with enough precision to be testable for conformance |
| 02:01 | <crocket> | How can I explain that to others? |
| 02:02 | <Philip`> | What are you trying to convince them of? |
| 02:02 | <crocket> | I need to explain that a web browser is better at displaying HTML4 than others. |
| 02:03 | <crocket> | I have to find the web browser that renders HTML4 best. |
| 02:03 | <crocket> | But |
| 02:03 | <crocket> | HTML4 was last published in Dec 1999. |
| 02:04 | <crocket> | I guess all web browsers don't really differ in their capacity to render HTML4. |
| 02:04 | <crocket> | Philip`, How do you think? |
| 02:05 | <crocket> | nothing? |
| 02:05 | <Philip`> | A web browser that implements what HTML4 specified will be worse in any real-world situations than one that doesn't (but that implements what HTML5 specifies instead) |
| 02:05 | <crocket> | huh? |
| 02:05 | <Philip`> | so HTML4 conformance is not a desirable quality (nor a well-defined one) |
| 02:05 | <crocket> | How about HTML4 rendering capacity? |
| 02:05 | <Philip`> | HTML4 defines lots of stuff that is incompatible with how real web pages work |
| 02:06 | <Philip`> | so browsers don't try to implement it |
| 02:06 | <crocket> | What the hell |
| 02:06 | <crocket> | What did the browsers render actually? |
| 02:06 | <crocket> | If it wasn't HTML4, what was it? |
| 02:06 | <zewt> | magic |
| 02:06 | <Philip`> | It was much more like what HTML5 now specifies |
| 02:06 | <crocket> | ok |
| 02:07 | <Philip`> | (since HTML5 is largely based on reverse-engineering what web browsers did) |
| 02:07 | <crocket> | Things were close to HTML5 since 10 years ago |
| 02:07 | <crocket> | Philip`, Are you a member of whatwg? |
| 02:08 | <Philip`> | "member" doesn't really mean anything in the WHATWG |
| 02:08 | <crocket> | Philip`, Have you participated in whatwg activities? |
| 02:08 | <Philip`> | Anyone can join the mailing list or IRC etc and join discussions |
| 02:08 | <Philip`> | Yes |
| 02:08 | <crocket> | IRC and mailing list are everything in whatwg? |
| 02:09 | <erlehmann> | crocket, the WHATWG is an adhocracy, only formally governed by the surpreme leader. |
| 02:09 | <Philip`> | That's where almost all communication occurs |
| 02:09 | <crocket> | ok |
| 02:09 | <erlehmann> | IRC is for tasteless jokes and wonkery, mailing list for tasteless specs and wonkery. |
| 02:09 | <crocket> | ha |
| 02:10 | <erlehmann> | i have no idea what wonkery means, btw. |
| 02:10 | <erlehmann> | but it sounds nice. |
| 02:10 | <erlehmann> | let me look it up |
| 02:10 | <crocket> | I already looked it up |
| 02:11 | <erlehmann> | wonkery (uncountable) |
| 02:11 | <erlehmann> | The quality or activities associated with being a wonk |
| 02:11 | <erlehmann> | sounds. nice. |
| 02:11 | Philip` | hopes our supreme leader is careful to avoid over-work while on trains |
| 02:11 | <erlehmann> | wonk (plural wonks) |
| 02:11 | <erlehmann> | (derogatory) An overly studious person, particularly student; a nerd. |
| 02:11 | <erlehmann> | (by extension) A policy wonk or other intellectual expert. |
| 02:11 | <erlehmann> | lol |
| 02:11 | <erlehmann> | that seems exactly right. |
| 02:12 | <erlehmann> | TabAtkins, your explanation on premultiplied color spaces was fine, btw. i had to dig it out again to show someone four-dimensional thinking ^_^ |
| 02:12 | <crocket> | overwork? |
| 02:12 | <crocket> | Does the leader have a job apart from being a leader? |
| 02:14 | <crocket> | Does whatwg make HTML standard? |
| 02:16 | <crocket> | "The WHATWG has a small, invitation-only steering committee called "Members", which has the power to impeach the editor of the specifications." |
| 02:16 | <crocket> | Which group among whatwg and W3C made HTML5? |
| 02:16 | <Philip`> | crocket: The editor is employed by Google to work on this stuff |
| 02:18 | <Philip`> | The "Members" have never done anything (since the editor hasn't needed impeaching yet) so they're not a particularly important group |
| 02:19 | <Philip`> | Have you seen http://www.whatwg.org/specs/web-apps/current-work/multipage/introduction.html#history-1 ? |
| 02:19 | <crocket> | Philip`, Did W3C participate in making HTML5? |
| 02:21 | <Philip`> | Yes, a few years after the WHATWG started work on it |
| 02:23 | <crocket> | ok |
| 02:23 | <crocket> | I guess whatwg is the main contributor to HTML5. |
| 02:23 | <crocket> | W3C wasn't really interested in HTML5. |
| 02:24 | <crocket> | ok |
| 02:24 | <crocket> | Now they are working together. |
| 02:25 | <Philip`> | The main contributors to HTML5 are individuals, who either use the WHATWG processes (mailing list and IRC etc) or W3C processes (mailing list and bug tracker and issue tracker and decision process etc) or both |
| 02:26 | <crocket> | thanks |
| 02:26 | <Philip`> | (The organisations don't really do anything by themselves, they just provide methods of communication) |
| 02:27 | <crocket> | ok |
| 02:28 | <crocket> | So |
| 02:28 | <crocket> | people made HTML5 via whatwg and W3C chanels |
| 02:35 | <jamesr_> | some organizations provide money to individuals in exchange for them working on specs |
| 03:45 | <Yuhong> | On quirks mode: http://www.reddit.com/r/web_design/comments/njb2r/im_a_newbie_and_made_a_website_that_i_think_is/ |
| 08:20 | <hsivonen> | hmm. Microsoft seems to be paying to advertise on the StackOverflow html5 tag |
| 08:44 | <bga> | http://evilbrainjono.net/pages/startup-or-pokemon.py |
| 09:06 | <annevk> | https://bugzilla.mozilla.org/show_bug.cgi?id=600715 oh sweet, encoding and decoding are different |
| 09:49 | <zcorpan> | heycam: for nullable types, i take it that null is considered to match the type in the overload resolution algorithm |
| 09:50 | <heycam> | zcorpan, ah yeah, I didn't mention nullable types did I |
| 09:51 | <heycam> | zcorpan, I would start assuming that it matches in the same way as currently and see if that works :) |
| 09:51 | heycam | will verify tomorrow |
| 09:53 | <zcorpan> | i'll reply on the list |
| 09:53 | heycam | heads off now |
| 10:05 | <zcorpan> | come to think of it, there was an example already with the null problem (options.add) |
| 10:05 | <annevk> | http://dvcs.w3.org/hg/encoding/raw-file/tip/Overview.html#encodings |
| 10:06 | <annevk> | "The table below lists all encodings and their labels user agents must support. User agents must not support any other encodings or labels." |
| 10:06 | <annevk> | also defines "get an encoding" now |
| 10:07 | <zcorpan> | "The Encoding"? |
| 10:08 | <annevk> | more like: http://krijnhoetmer.nl/irc-logs/whatwg/20111219#l-507 |
| 10:09 | <annevk> | couldn't really think of a better title |
| 10:13 | <annevk> | so basically all ISO-2022 encodings are as dangerous as UTF-7?! |
| 10:18 | <annevk> | zcorpan: why does null matter? |
| 10:19 | <zcorpan> | annevk: the overload resolution ends up with several candidates if null matches |
| 10:20 | <zcorpan> | and we clearly need null to match, or nullable types can't be used with overload |
| 10:21 | <annevk> | I think picking one of them at random should always yield the same outcome |
| 10:47 | <annevk> | anyone time to review an email for me? |
| 10:52 | <zcorpan> | is it long? |
| 10:53 | <annevk> | 5 short paragraphs |
| 10:53 | <zcorpan> | then sure |
| 10:53 | <annevk> | just want to know if I should include more info or if it's okay |
| 10:53 | <annevk> | you got it |
| 10:55 | <zcorpan> | looks good |
| 10:56 | <annevk> | thanks |
| 11:24 | <annevk> | argh |
| 11:24 | <annevk> | try to contribute to MDN |
| 11:24 | <Ms2ger> | MikeSmith, can has a webapps testsuite component in bugzilla? |
| 11:25 | <annevk> | recover password, try to edit profile to change it, starts redirecting me to some /nl/ URL |
| 11:26 | <Ms2ger> | We all hate it :) |
| 11:27 | <annevk> | how does it even know I want a /nl/ profile? |
| 11:27 | <annevk> | I only contributed to /en/ |
| 11:27 | <Ms2ger> | And somebody remind me later that I fix the WebSocket testsuite for Mozilla's unprefixing? |
| 11:27 | <annevk> | bah |
| 11:28 | <annevk> | still no idea on how to change the password |
| 11:28 | <annevk> | :/ |
| 11:44 | <hsivonen> | hmm. looks like the ICU team at IBM discovered HTML5. calendar systems ahead. |
| 11:45 | <annevk> | what do you mean? |
| 11:47 | <hsivonen> | annevk: it seems that they don't believe in the Gregorian calendar being the only one relevant to the use cases for calendar <input>s |
| 11:48 | <annevk> | UI or submission? |
| 11:48 | <hsivonen> | we'll see |
| 11:48 | <annevk> | is there some discussion somewhere? |
| 11:49 | <hsivonen> | annevk: https://www.w3.org/Bugs/Public/show_bug.cgi?id=15278 |
| 11:50 | <hsivonen> | comes with a Word file that suggests adding new input types |
| 11:53 | <annevk> | and the word "business needs" as use case |
| 11:53 | <annevk> | s* |
| 11:53 | <hsivonen> | annevk: that's a very specific use case |
| 11:55 | <hsivonen> | How did the Web Socket API manage to get to CR already at the W3C? |
| 11:58 | <annevk> | just like that |
| 11:59 | <jgraham> | Is there a testsuite? |
| 11:59 | <jgraham> | I know one isn't needed for CR… |
| 12:01 | <annevk> | I think some people submitted tests |
| 12:02 | <hsivonen> | if a test suite isn't needed for CR, we should take more things to CR in order to get them unprefixed |
| 12:03 | <jgraham> | Per the Process a testsuite is only neede for PR I think |
| 12:03 | <jgraham> | Since you have to demonstrate interoperability |
| 12:03 | <jgraham> | Of course the Process is a lie |
| 12:04 | <annevk> | it's a cake! |
| 12:05 | <annevk> | hsivonen: lots of DOM/XHR stuff goes in without prefix |
| 12:05 | <annevk> | also HTML stuff |
| 12:05 | <annevk> | it's just CSS that has the dated policy |
| 12:05 | <annevk> | and frankly harmful to non-WebKit browsers |
| 12:06 | <hsivonen> | hmm. TAG help sought for fixing the broken CA system |
| 12:10 | <annevk> | hsivonen: brucel wonders if validator.nu supports GET requests for Text field input |
| 12:13 | <hsivonen> | annevk: It doesn't |
| 12:13 | <hsivonen> | annevk: it support GET requests with the input document as a data: URL |
| 12:18 | <zcorpan> | hmm. my w3c-bug-links.js userjs broke. |
| 12:26 | <hsivonen> | sigh. the TAG minutes refer to Crockford on HTML and Security |
| 12:27 | <annevk> | run for the TAG next year? |
| 12:27 | <annevk> | I have contemplated it just to give it a chance (if I manage to get elected that is) |
| 12:29 | <wilhelm> | What's the worst thing that can happen? Go for it. (c: |
| 12:29 | <jgraham> | It is not clear that having a TAG with more of an on-the-ground experience of the web as it is understood by browser makers/users would be a bad thing |
| 12:30 | <jgraham> | wilhelm: I imagine the worst thing that can happen is that timbl turns into a zombie and eats your brains. Second worst is probably years of frustrating and ineffectual meetings |
| 12:30 | <annevk> | I am mostly worried about second worst |
| 12:31 | <jgraham> | Right, that was ordered by absolute badness not by expected badness (i.e. absolute badness of event x risk of event happening) |
| 12:57 | MikeSmith | looks around for Ms2ger |
| 12:57 | <MikeSmith> | Test suite component added for Webapps WG https://www.w3.org/Bugs/Public/describecomponents.cgi?product=WebAppsWG |
| 12:58 | <hsivonen> | http://lists.w3.org/Archives/Public/wai-xtech/2011Dec/0008.html |
| 13:00 | <Ms2ger> | MikeSmith, thanks |
| 13:13 | <MikeSmith> | annevk: you can get MDN help on #devmo on irc.freenode.net |
| 13:13 | <Ms2ger> | On irc.mozilla.org |
| 13:33 | <annevk> | http://dump.testsuite.org/encoding/label-test.html |
| 14:15 | <hsivonen> | aargh. UMP still showing up in a thread about advancing CORS |
| 14:17 | <MikeSmith> | yeah |
| 14:46 | <annevk> | so error handling does indeed differ in decoding shift_jis |
| 14:47 | <annevk> | e.g. for bytes in the series lead byte 84, second byte 00-FF, WebKit/Gecko will treat each byte individually until the second byte hits 40, whereas Opera always treat two octets identical when there's a lead byte |
| 14:49 | <annevk> | I mean Opera treats them as one, and as such 84 33 for instance gets turned into U+FFFD rather than U+FFFD U+0033 |
| 14:49 | <annevk> | I wonder what IE does |
| 14:56 | <hsivonen> | it's so sad every time that someone says that fingerprinting isn't a problem because the user can modify he UA string |
| 14:56 | <annevk> | reminds of "UI isn't a problem, you can customize it!" |
| 14:57 | <zewt> | it's not like it'll ever be possible to mask which browser you're using if scripting is enabled. heh |
| 14:57 | <hsivonen> | annevk: except in the fingerprinting case, saying that you can modify it is missing the point even more |
| 14:57 | <annevk> | but I guess you mean that changing the UA string will make it worse |
| 14:57 | <hsivonen> | annevk: yes |
| 14:58 | <jgraham> | Much worse I presume |
| 14:58 | <Ms2ger> | hsivonen, you can just submit the reftests, the tools will save us later |
| 14:58 | <jgraham> | The number of people running something that looks like foo browser with a bar browser UA string is tiny |
| 14:59 | <jgraham> | So you can fingerprint people almost uniquely just from that |
| 14:59 | <hsivonen> | jgraham: righht |
| 14:59 | <jgraham> | Ms2ger, hsivonen: what tests? |
| 15:00 | <hsivonen> | Ms2ger: do I just put the files and their references to the submission directory and leave them there until the tools save us |
| 15:00 | <Ms2ger> | jgraham, document.write, p-html-testsuite |
| 15:00 | <Ms2ger> | hsivonen, yeah |
| 15:00 | <hsivonen> | jgraham: http://lists.w3.org/Archives/Public/public-html-testsuite/2011Dec/0008.html |
| 15:01 | <jgraham> | Oh, I have email :) |
| 15:01 | <hsivonen> | jgraham: I'm mainly hoping that Kris forwards that to their HTML parser person, because I have no faith in Microsoft's official bug report channel |
| 15:02 | <hsivonen> | jgraham: I figured you'd see it as a side effect so I don't need to file a bug on Opera |
| 15:02 | <jgraham> | hsivonen: Some style points: could you rename bug{bmo-id}.js to something else? Can you make the expected text the word PASS or similar? |
| 15:03 | <jgraham> | I haven't looked too closely to see if the second part is practical) |
| 15:03 | <hsivonen> | jgraham: I need a 6-letter synonym for pass that does not have duplicate letters |
| 15:04 | <zewt> | some kind of crossword? heh |
| 15:04 | <zewt> | SUPERB |
| 15:04 | <jgraham> | hsivonen: At the very least put in the test file "you should see the string "CcBbAa" below" |
| 15:04 | <hsivonen> | zewt: thanks |
| 15:04 | <hsivonen> | jgraham: ok |
| 15:05 | jgraham | wonders how "SUPERB" is a synonym of PASS |
| 15:05 | <jgraham> | hsivonen: Thanks |
| 15:06 | <zewt> | passing is superb! |
| 15:06 | <zewt> | failing not so much |
| 15:06 | <zcorpan> | ravine |
| 15:07 | <jgraham> | zcorpan: Isn't the ravine waht you avoid by taking the pass? So ravine is more like fail :p |
| 15:07 | <zewt> | これは正解だ |
| 15:08 | <zcorpan> | plight |
| 15:08 | <zcorpan> | jgraham: i'm just checking my synonym dictionary :) |
| 15:11 | <jgraham> | zcorpan: It sounds more like an atonyms dictionary :) |
| 15:12 | <zcorpan> | cruise depart linger travel answer muster convey demise depart vanish ordain ratify |
| 15:12 | <Ms2ger> | I guess depart works |
| 15:13 | <zcorpan> | oh i got depart twice |
| 15:13 | <zewt> | doesn't sound like "pass" to me |
| 15:13 | <jgraham> | I think I have just learnt that there are no six letter synonyms for the correct sense of pass that don't repeat letters |
| 15:13 | <jgraham> | English needs more words |
| 15:14 | <zcorpan> | zewt: it was the most common synonym that matched the constraints for all meanings of 'pass' in my dictionary! |
| 15:14 | <jgraham> | zewt: It does match pass as in "passed out of town" |
| 15:14 | <zewt> | i've never heard "pass" used that way |
| 15:14 | <zewt> | the only meaning I can think of for both is "died" |
| 15:14 | <jgraham> | or "passed away" |
| 15:15 | <zewt> | which I suppose would be an entertaining way of reading test reports |
| 15:15 | <zewt> | congratulations, all of your tests died |
| 15:15 | <zcorpan> | now i should go to the gym before i pass |
| 15:15 | <zcorpan> | sorry, i should pass to the gym before i pass |
| 15:16 | <Philip`> | "alrite" |
| 15:16 | <zewt> | ew |
| 15:16 | <Philip`> | "worked" |
| 15:17 | <Philip`> | "IsOkay" |
| 15:17 | <zcorpan> | is mixing uppercase and lowercase ok? pasSed |
| 15:18 | <jgraham> | PaSsEd is quite nice |
| 15:18 | <hsivonen> | "worked" works |
| 15:18 | <Philip`> | pas5ed |
| 15:18 | <hsivonen> | thanks |
| 15:19 | <zcorpan> | do you need one for "fail" too? "fucked"? |
| 15:19 | zcorpan | really goes now |
| 15:20 | <annevk> | I wonder if this is a serious problem |
| 15:20 | <Philip`> | "failed" would probably cause less offence, and therefore be inferior |
| 15:20 | <annevk> | in shift_jis |
| 15:20 | <annevk> | 84 3C 73 63 72 69 70 74 20 84 3E |
| 15:21 | <aleph> | Hi, why in HTML5 http-equiv value is case sensitive? IN HTTP headers it case-insensitive! Example 'content-type' |
| 15:21 | <annevk> | gives <script> in Gecko/Chrome, but "�script �" in Opera |
| 15:21 | <annevk> | abarth: ^^ |
| 15:24 | <annevk> | aleph: it's case-insensitive |
| 15:24 | <aleph> | http://bit.ly/shbUFq |
| 15:24 | <annevk> | aleph: see http://www.whatwg.org/C#enumerated-attribute |
| 15:24 | <aleph> | well, in HTML5 spec, it says for example, "content-type", and doesn't say it is case insensitive, so "Content-Type" becomes invalid |
| 15:24 | <annevk> | aleph: I gave you a pointer to the HTML spec, where it says it's an enumerated attribute, which are case-insensitive |
| 15:24 | <annevk> | aleph: the document you pointed to is non-normative and might require some clarification |
| 15:25 | <zewt> | annevk: i'd expect the only safe way to output user-entered text in a particular encoding is to ensure that it's a valid sequence in that encoding |
| 15:26 | <zewt> | (eg. convert to UTF-8 then back to the source encoding, or something like that) |
| 15:26 | <aleph> | annevk: Thanks you! :) |
| 15:26 | <aleph> | everything is clear now |
| 15:27 | <annevk> | zewt: sure, but that wasn't really what I was asking for |
| 15:27 | <zewt> | i wonder if anyone really is trying to do escaping without that encoding normalization/sanitization first |
| 15:28 | <annevk> | well an attacker |
| 15:28 | <Philip`> | annevk: I expect anything that's attempting to output safely-escaped text will either work on bytes, so 0x84 0x3C will become 0x84 "<" which is hopefully safe, or else will decode to Unicode characters then escape then re-encode (giving validly-encoded output) |
| 15:29 | <annevk> | https://bugzilla.mozilla.org/show_bug.cgi?id=593338 explains the middle dot in Gecko, seems hsivonen ran into it too |
| 15:30 | <zewt> | Philip`: the spec allows multibyte encodings where < appears in the middle of a multibyte sequence, though, so escaping bytewise would break text |
| 15:30 | <Philip`> | Seems very unlikely anyone would e.g. decode the user input, check the decoded version has no forbidden characters, then output the original user input (so it may have forbidden characters when decoded differently) |
| 15:30 | <zewt> | (i wish that could be prohibited, but iso-2022 ...) |
| 15:31 | <zewt> | (iso-2022 is more evil than anything utf-16 could ever contemplate) |
| 15:31 | <Philip`> | zewt: That's just an encoding bug, which is not a security problem - if people want to fix that bug they'll decode to Unicode before escaping and it'll be fine |
| 15:32 | <annevk> | Philip`: you could have a cp932 pipeline that goes through the incoming bytes, skips bytes after lead byte and checks everything else, and outputs straight away |
| 15:32 | <zewt> | Philip`: i mean, doing it bytewise is just an incorrect way to do it (if you want to support generic encodings); you have to use the convert-to-unicode method |
| 15:33 | <hsivonen> | http://w3c-test.org/html/tests/submission/Mozilla/nested-document-write-1.html |
| 15:33 | <hsivonen> | http://w3c-test.org/html/tests/submission/Mozilla/nested-document-write-2.html |
| 15:35 | <Ms2ger> | Thanks |
| 15:36 | <jgraham> | hsivonen: Thanks |
| 15:36 | <zewt> | are there any multibyte encodings that alias the ASCII range (especially < or %) other than iso-2022? |
| 15:36 | <zewt> | i'm not familiar with chinese and korean encodings |
| 15:37 | <annevk> | there was utf-7 which got killed |
| 15:37 | <annevk> | for this very reason |
| 15:37 | <zewt> | well plus "nobody anywhere uses it" ... which unfortunately isn't the case with iso-2022 :( |
| 15:38 | <zewt> | pretty sure i've seen iso-2022-jp in the wild... |
| 15:38 | <annevk> | damnit I need IE if I want to test cp932 properly |
| 15:39 | <zewt> | i wonder if most iso-2022 pages only change the upper mapping and leave the low ASCII-overlap one alone; if "iso-2022 for the web" could be profiled to require that, it would at least put a dent in the evilness (and the security issues) |
| 15:39 | <zewt> | (eg. ignore requests to change the "GL" mapping) |
| 15:40 | <Philip`> | zewt: JOHAB, I think |
| 15:40 | <zewt> | wow. never even seen that sequence of letters before |
| 15:40 | <jgraham> | hsivonen: I wasn't expecting you to not submit a ref at all |
| 15:40 | <Philip`> | (plus obviously EBCDICs but they're not multi-byte) |
| 15:40 | <annevk> | zewt: see http://coq.no/character-tables/mime/iso-2022/en for some info |
| 15:41 | jgraham | actually has some similar tests somewhere |
| 15:41 | <Philip`> | (I think the list near the end of http://www.whatwg.org/specs/web-apps/current-work/multipage/semantics.html#charset is reasonably comprehensive) |
| 15:44 | <annevk> | Gecko supports x-johab according to Web Encodings... |
| 15:45 | <zewt> | at least 2022 doesn't allow setting GR to ASCII, apparently |
| 15:55 | <Philip`> | It's unfortunate that with all of these ASCII-conflicting encodings, they're mainly a security problem when the server supports the encoding and the browser doesn't |
| 15:55 | <Philip`> | and it's much easier to update browsers than servers |
| 15:55 | <Philip`> | so the result is more encoding proliferation |
| 15:56 | <Philip`> | (and even if nobody actually uses the encodings in practice, browsers can't remove them without introducing XSS vulnerabilities on other people's sites) |
| 15:57 | <annevk> | can you actually do something sensible with the page though if you cannot read it? |
| 15:57 | Philip` | doesn't understand the question |
| 15:58 | <annevk> | maybe I don't understand the attack scenario |
| 15:59 | <annevk> | actually, nm |
| 15:59 | <annevk> | I do now :) |
| 15:59 | Philip` | is thinking of the scenario in http://zaynar.co.uk/docs/charset-encoding-xss.html |
| 16:03 | <annevk> | you got a home page now? :) |
| 16:03 | annevk | only knew about http://philip.html5.org/ |
| 16:03 | <Philip`> | (Chrome added support for iso-2022-kr after Yahoo notified them that it triggered the vulnerability in Yahoo search) |
| 16:04 | <Philip`> | It's not really a home page; it's just another page in a random location with a random collection of stuff on it :-p |
| 16:04 | <divya> | Philip`: is this tweetableeee |
| 16:05 | <annevk> | Philip`: nice article btw |
| 16:05 | <Philip`> | divya: If that word means what I think it means, then I guess so |
| 16:06 | <divya> | yes it does! |
| 16:06 | <annevk> | Philip`: and too bad about Chrome, what's next, EBCDIC? |
| 16:06 | <divya> | thnx |
| 16:06 | <Philip`> | (though it is a couple of years old now and the Ultraseek example doesn't work because Ultraseek died) |
| 16:07 | <divya> | :)) |
| 16:07 | <divya> | Philip`: are you on twitter? |
| 16:08 | <beverloo> | @foolip |
| 16:08 | <annevk> | no that's a different philip |
| 16:09 | <divya> | for shame beverloo |
| 16:09 | <beverloo> | ah, then I've been mixing them up for a while now :-) |
| 16:09 | <beverloo> | sorry |
| 16:09 | <annevk> | this is Philip lowercase Taylor |
| 16:09 | <annevk> | the other is Jägenstedt |
| 16:09 | <Philip`> | Hmm, I remember finding the same bug in Google Groups later, but Chrome supported iso-2022-kr by that point so it could only trigger XSS in Konqueror |
| 16:09 | Philip` | had forgotten about that |
| 16:09 | <annevk> | there's also Philip TAYLOR, but I haven't seen him in a while |
| 16:10 | <Philip`> | divya: I'm not |
| 16:10 | <jgraham> | Pretty sure Philip` isn't on twitter. Unless he is a pro footballer in disguise |
| 16:10 | <divya> | hahahaha |
| 16:22 | <zewt> | Philip`: they could have (once upon a time) made it so scripts don't execute if the encoding isn't known; too late now... |
| 16:22 | <zewt> | or just blacklist scripts for known problem encodings like 2022 |
| 16:23 | <zewt> | (if the <script> security bug was really the only thing that made them implement it) |
| 16:27 | <Philip`> | zewt: That would confuse people who write "content-type: text/html;charset=iso-8559-1" (sic) and find that everything looks fine but their scripts stop working, which may not be ideal |
| 16:27 | <zewt> | Philip`: googlable confusion would have been an improvement over this mess |
| 16:28 | <zewt> | maybe hard to google, i guess; still better |
| 16:28 | <zewt> | (and I guess "googlable" may be an anachronism for when this problem really started) |
| 16:28 | <zewt> | Philip`: but that's also fixable |
| 16:28 | <Philip`> | Blacklisting just 2022/etc could work, I suppose - if Chrome got away with not supporting iso-2022-kr for years then presumably it's not needed for compatibility, and they could decide to decode it bogusly and disable scripts (or pretend it's text/plain or whatever), instead of using a proper 2022 decoder |
| 16:28 | <zewt> | it wouldn't have disabled scripts for all unrecognized encodings; just for an known blacklist |
| 16:29 | <zewt> | (sorry, off writing an email and losing my train of thought here; damn you, day job) |
| 16:31 | <annevk> | you cannot really say much about iso-2022-kr because that market used to be predominantly IE |
| 16:31 | <annevk> | and still is probably |
| 16:31 | <annevk> | so other browsers get likely way less feedback |
| 16:32 | <Philip`> | zewt: If we can go back in time, I suppose we should just require HTML to be UTF-8 and not accept anything else |
| 16:32 | <zewt> | i'm still curious whether Firefox defaulting to Japanese for CJK is really okay, or if it's just a lack of C/K feedback |
| 16:32 | <zewt> | that makes it look okay |
| 16:32 | <zewt> | Philip`: so many things we could do if we could do that :) |
| 16:33 | <dglazkov> | good morning, Whatwg! |
| 16:38 | <kennyluck> | zewt, there are Mozilla communities in all three C, J and K so I don't think lack of feedback is an issue. |
| 16:38 | <Philip`> | zewt: Hmm, yeah, if we could go back in time I suppose we should prioritise things like killing Hitler over improving the HTML character encoding situation |
| 16:38 | <kennyluck> | FWIW, zh-tw ships UTF-8 as the default language regardless of what the spec says and what IE does… |
| 16:38 | <zewt> | encoding |
| 16:38 | <kennyluck> | which was quite shocking to me when I heard this. |
| 16:39 | <zewt> | kennyluck: i hope you're right; it just seems hard to believe that nobody in China would complain about fonts defaulting to Japanese |
| 16:39 | <kennyluck> | zewt, yeah *encoding. |
| 16:39 | <zewt> | (I'm not objecting--I'd be very happy if all browsers would do that) |
| 16:40 | <zewt> | just pessimisting :) |
| 16:40 | <kennyluck> | zewt, Firefox has marginal market share in China, so. |
| 16:43 | <zewt> | kennyluck: seems like you're pointing in two directions, so I'm confused |
| 16:43 | <zewt> | either defaulting to Japanese is okay, it's not okay (but nobody's getting feedback complaining about it), or ... something else that I havn't thought of |
| 16:44 | <kennyluck> | zewt, you are right. |
| 16:44 | <zewt> | which *should* be largely a question of existing content, though politics could enter into it as well |
| 16:45 | <kennyluck> | It's OK to me. (it's just not distinguishable to me). I can't speak for people who are picky though I know whom to ask this question. |
| 16:46 | <zewt> | my sense (and it's just a vague sense; I'm neither C nor J) is that Chinese people are less sensitive to the font issue than Japanese; so Japanese fonts are a less objectionable default (even when it's wrong for existing content that lacks @lang) |
| 16:48 | <kennyluck> | zewt, let me ask again. You are referring to this difference right? http://en.wikipedia.org/wiki/Han_unification#Examples_of_language_dependent_characters |
| 16:48 | <kennyluck> | zewt, that's possible. |
| 16:49 | <zewt> | kennyluck: well, that's the underlying problem, but what I"m really asking about is whether Firefox's particular behavior is okay and could be (maybe, hopefully, with some pushing and convincing) done by other browsers |
| 16:49 | <zewt> | which is: no language heuristics (as Opera does), no locale sensitivity (as IE does); if text is UTF-8 and has no @lang, CJK just uses Japanese fonts; no magic |
| 16:50 | <zewt> | i had just automatically assumed that sort of thing would never be accepted, until I found out that Firefox apparently does just that |
| 16:52 | <kennyluck> | zewt, I guess I can ask people. Although that's really less a problem to me. Do you have test cases for this? |
| 16:53 | <zewt> | http://zewt.org/~glenn/encoding%20utf-8.html |
| 16:53 | <kennyluck> | or perhaps I should just hack the table in the wiki to include one for @lang=en |
| 16:53 | <zewt> | kennyluck: note that it's actually a little more complicated than that |
| 16:53 | <zewt> | since you can have CJK characters inside lang=en |
| 16:53 | <zewt> | in which case you again have to pick a default somehow |
| 16:54 | <zewt> | hold on I'll add a lang=en to that |
| 16:54 | <zewt> | just for completeness |
| 16:54 | <kennyluck> | Does @lang="" gives you the default or it's ignored? |
| 16:54 | kennyluck | is checking the spec |
| 16:54 | <zewt> | i havn't tested what an explicitly blank @lang does (I forget if that's even allowed) |
| 16:56 | <zewt> | added en-US (and zh-CN, while I was there) |
| 16:57 | <annevk> | basing in-page UI on the user's locale is an antipattern Lachy |
| 16:57 | <annevk> | I wonder if someone has written an article on that yet |
| 16:58 | <zewt> | annevk: sure, but that doesn't make IE stop being IE :( |
| 16:58 | <zewt> | heuristics are also bad (but less bad, since at least in *theory* they could be made interoperable) |
| 16:58 | <annevk> | wait what? |
| 16:59 | <annevk> | was that in reply to in-page UI? |
| 16:59 | <annevk> | or the stuff about Korea? |
| 16:59 | <zewt> | i mean, IE's default languages and encodings depends on the user's locale |
| 16:59 | <zewt> | depend |
| 16:59 | <annevk> | oh, I was talking about stuff like <input type=date> |
| 17:00 | <annevk> | or e.g. data:text/html,<input type=submit> |
| 17:00 | <zewt> | in-page UI seems okay *if* it has no detectable effect on scripts |
| 17:01 | <zewt> | (or submissions to the server, etc., of course) |
| 17:35 | <annevk> | inbox < 400 \o/ |
| 17:37 | <jgraham> | Putting productivity gurus to shame there |
| 17:37 | <annevk> | MikeSmith: not sure if you're around, but can we merge XHR 1.0 and XHR 2.0 in Bugzilla? (just XHR will do) |
| 17:38 | <annevk> | smaug____: has Gecko disabled cross-origin synchronous requests already for XMLHttpRequest? |
| 17:39 | <annevk> | jgraham: it's going fairly well I think, I haven't addressed this many problems in quite a while |
| 17:39 | <annevk> | and I'm generating a new class of problems as I go of course (encodings aaaah) |
| 17:40 | <kennyluck> | zewt, did you ever test Firefox on Windows about this CJK glyph issue? I got some… madness |
| 17:40 | <zewt> | i'm in windows |
| 17:40 | <zewt> | can you narrow "madness"? :P |
| 17:40 | <zewt> | (the entire topic is madness) |
| 17:40 | <smaug____> | annevk: looking... |
| 17:40 | <kennyluck> | So let me say this. I have a zh-TW Windows, specifically Windows 7. |
| 17:41 | <smaug____> | annevk: er, hmm, I don't think cors has been disabled yet |
| 17:41 | <zewt> | i'm in win7, set to JP; also tested in a ZH codepage (forget whether it was TW or CN, TW I think) |
| 17:42 | <kennyluck> | For zh-TW FF nightly, the default @lang gives the glyph for zh-TW and zh-CN (they are the same), so does @lang=en. The @lang=ja one is different. |
| 17:42 | <annevk> | ooh you're on windows? |
| 17:42 | <smaug____> | annevk: actually I didn't remember that cors should be disabled |
| 17:42 | <zewt> | zh-TW and zh-CN should not be the same (they're very visually discernable) |
| 17:42 | <smaug____> | that is a bit risky |
| 17:42 | <smaug____> | comparing to other changes |
| 17:42 | <smaug____> | cors is old stuff |
| 17:43 | <zewt> | testing in FF8 (don't know if this has changed since nightly) |
| 17:43 | <kennyluck> | For ja FF nightly (on a zh-TW Windows, remember), the default @gives the glyph for zh-TW and the glyph for zh-CN is different from the one for zh-TW. And @lang=en gives the @lang-ja glyph, which is another glyph. |
| 17:43 | <zewt> | i'm going to have to draw a chart to figure out all the things you just said :) |
| 17:44 | <zewt> | (first impression: i agree with "madness") |
| 17:44 | <zewt> | http://zewt.org/~glenn/encoding%20expected.png is what I see |
| 17:44 | <annevk> | smaug____: yeah, sicking wanted that |
| 17:44 | <kennyluck> | But in any case, OS-dependent seems weird. |
| 17:44 | <zewt> | and I think the jp/tw/cn glyph choices are correct |
| 17:44 | <zewt> | in that png |
| 17:45 | <annevk> | smaug____: and you said it was risky then too http://www.w3.org/Bugs/Public/show_bug.cgi?id=14773 |
| 17:45 | <zewt> | ... though my impression of "correct" may have come from incorrect browsers, so I should recheck |
| 17:45 | <zewt> | err, hold on--what does "ja FF nightly" mean? |
| 17:46 | <kennyluck> | zewt, firefox-11.0a1-ja.win32.install.exe is what I installed on a zh-TW Windows7. |
| 17:47 | <zewt> | why would there be different language installers? that's insane. |
| 17:47 | <zewt> | and having that affect rendering: even more insane |
| 17:47 | <kennyluck> | zewt, so when you say locale-dependent. You mean OS-locale? |
| 17:48 | <zewt> | in the case of IE, yes |
| 17:48 | <kennyluck> | hmm…. OK. |
| 17:48 | <zewt> | i had no idea Firefox had dependencies on which language installer you used |
| 17:48 | <zewt> | if true, i lose much hope for humanity's future |
| 17:50 | <zewt> | (all testing I've done has been with the default FF8 installer, "Firefox Setup 8.0.1.exe" (or maybe 8.0.0 at the time) |
| 17:50 | <kennyluck> | zewt, I didn't set OS-locale as you did, so. But the normal zh-TW on a zh-TW machine is respecting the locale… |
| 17:50 | <jgraham> | Does anyone know someone that works on jQuery? |
| 17:51 | <jgraham> | Their documentation seems to be broken |
| 17:51 | <kennyluck> | for defalult @lang |
| 17:51 | <jgraham> | Everything is uncategorised |
| 17:51 | <kennyluck> | and not giving @lang=ja glyph. |
| 17:51 | <zewt> | let me figure out which VM I was using |
| 17:54 | <zewt> | kennyluck: fwiw, also http://krijnhoetmer.nl/irc-logs/whatwg/20111207#l-575 |
| 17:58 | <zewt> | kennyluck: oddly, in FF8 in XP in zh-TW, I don't see the TW glyph at all (zh-TW shows the JP glyph) ... maybe XP's fonts didn't have that variant |
| 17:59 | <zewt> | loading clunky win7 vm so I can change codepages... |
| 18:03 | kennyluck | is now sending pictures to www-archive |
| 18:05 | <zewt> | i changed the locale, it goes "you have to restart", i click restart |
| 18:06 | <zewt> | it reboots. i log in. "now you have to restart" |
| 18:06 | <zewt> | rage. |
| 18:13 | <kennyluck> | zewt, I think hsivonen's statement is perhaps referring to the result on Mac OSX, which I can reproduce. |
| 18:13 | <zewt> | kennyluck: i get the same results in "Chiense (Traditional, Taiwan)" (CP950) as Japanese (CP932) in FF8.0.1 using Firefox Setup 8.0.1.exe |
| 18:13 | <zewt> | his statement holds in my testing, at least |
| 18:17 | <zewt> | is Firefox really doing the same horrible thing as IE (except using the installer as the locale selection, instead of the system codepage)? :| |
| 18:22 | <zewt> | if that's the case I'm probably just going to give up on this, since there's no sane precedent to move towards at all |
| 18:34 | <gsnedders> | Man, Peacekeeper is such a ridiculous, useless benchmark |
| 18:35 | <gsnedders> | (it tests everything in a loop: a lot of the tests are effectively no-ops after the first iteration) |
| 18:49 | <kennyluck> | zewt, ok, got some pictures on the Web finally → http://lists.w3.org/Archives/Public/www-archive/2011Dec/0024 |
| 19:04 | <zewt> | what is www-archive, exactly? the description on lists.w3.org is useless |
| 19:06 | <othermaciej> | zewt: it's for when you have an off-list message that you want to be archived and visible for the record, or to send random stuff simply to insert it into mailing list archives, or to complain about people supposedly privately but actually publicly |
| 19:07 | <zewt> | can I post on it without subscribing? because it sounds like a useless list to subscribe to, but I'm replying to a message CC'd there |
| 19:07 | <zewt> | (the "subscriber-post-only" vs. "spam" dilemma is endlessly annoying) |
| 19:07 | <kennyluck> | sane people won't subscribe to that list. |
| 19:08 | <zewt> | i don't want to, but lots of lists will bounce messages to non-subscribers (or annoy some moderator somewhere) |
| 19:08 | <zewt> | so ... just checking |
| 19:08 | <kennyluck> | so yeah, you can post on it without subscribing. |
| 19:09 | <kennyluck> | the first-time-manual check for W3C seems to work quite well. I don't know why though. |
| 19:37 | <MikeSmith> | Philip`: if you have time to look into https://www.w3.org/Bugs/Public/show_bug.cgi?id=12539 would appreciate it |
| 19:38 | <MikeSmith> | otherwise I will hack something into the script that generates the W3C version |
| 19:40 | <annevk> | hey MikeSmith |
| 19:40 | <MikeSmith> | hey |
| 19:40 | <annevk> | did you see my question about merging XHR 1.0 and 2.0 components? |
| 19:40 | <annevk> | can we rename components at all? |
| 19:41 | <MikeSmith> | yeah, we can rename them |
| 19:41 | <annevk> | MikeSmith: I could move all XHR 1.0 bugs to XHR 2.0 |
| 19:41 | <annevk> | and then we can rename XHR 2.0 and nuke XHR 1.0 |
| 19:41 | <MikeSmith> | yeah, could do that |
| 19:41 | <MikeSmith> | that's what I've done on some other cases |
| 19:42 | <annevk> | working on it |
| 19:43 | <annevk> | okay |
| 19:43 | <annevk> | MikeSmith: XHR 1.0 is empty and can be removed, XHR 2.0 can be renamed to XHR |
| 19:46 | <MikeSmith> | annevk: OK, done |
| 19:46 | <annevk> | thanks! |
| 19:46 | <annevk> | we can maybe do some more cleanup |
| 19:47 | annevk | checks how much work it would be to obsolete the DOM Range component |
| 19:48 | <annevk> | ok that looks like it might be worth it |
| 19:48 | <annevk> | MikeSmith: DOM Range can be removed too |
| 19:49 | <annevk> | and DOM Core could be named DOM I guess |
| 19:49 | <smaug____> | will fullscreen spec be done in webapps? |
| 19:50 | <smaug____> | if so, would be nice to get a component for it |
| 19:50 | <MikeSmith> | annevk: ok, done and done |
| 19:50 | <jamesr> | how do people typically check for concepts across the whatwg html spec (for example i want to see who, if anyone, defines pre-click activation behaviors). do you load up the single-page version and ctrl-f find, scrape a local copy, pull the source? |
| 19:51 | <annevk> | smaug____: dunno |
| 19:51 | <zewt> | i end up loading the full-page version (always after some hesitation, of course) |
| 19:51 | <zewt> | browser crush |
| 19:52 | <jamesr> | it's behaving pretty reasonably for me so far in chrome 16 |
| 19:52 | <annevk> | thanks again MikeSmith |
| 19:52 | <MikeSmith> | cheers |
| 19:52 | <zewt> | it still takes a few seconds to settle down ... everyone dreams of working xrefs in multipage, of course |
| 19:53 | <smaug____> | jamesr: usually I load the fullpage version (which takes time in all the browsers, but after loading behaves ok) and then ctrlf-f |
| 19:53 | <zewt> | chrome's doing a lot better than ff8, to be sure |
| 19:54 | <zewt> | at least it doesn't trigger the "slow, ain't it?" overlay |
| 19:54 | <annevk> | jamesr: full page, and you can click on the definitions then |
| 19:54 | <jamesr> | annevk: can click the definition, but you still have to ctrl-f to find references (like which elements define pre-click activation behaviors), right? |
| 19:54 | <annevk> | jamesr: e.g. if you click on "enumerated attributes" where it is defined it lists all the places that reference it |
| 19:55 | <jamesr> | ooh |
| 19:55 | <smaug____> | zewt: chrome does trigger the "slow" warning |
| 19:55 | <jamesr> | wait where can i see references? |
| 19:55 | <zewt> | smaug____: doesn't for me ... depends on your processor, of course |
| 19:55 | <annevk> | jamesr: click on the bold term |
| 19:55 | <smaug____> | zewt: and actually during slow one can't even really scroll the page on chrome, and resizing the window takes lots of time |
| 19:56 | <smaug____> | s/slow/load/ |
| 19:56 | <jamesr> | oh, the bold black non-underlined things are links? |
| 19:56 | <jamesr> | nice |
| 19:56 | <annevk> | they're definitions with special behavior |
| 19:56 | <zewt> | resizing (horizontally) takes about 1-2s for me |
| 19:56 | <smaug____> | which is a lot |
| 19:57 | <zewt> | jamesr: bold black and bold orange, at least |
| 19:57 | <zewt> | eg. "pre" in the section header in http://www.whatwg.org/specs/web-apps/current-work/#the-pre-element |
| 19:57 | <jamesr> | very interesting |
| 19:58 | <annevk> | http://dvcs.w3.org/hg/encoding/raw-file/tip/Overview.html has it too |
| 19:59 | <annevk> | e.g. you can click the encoding headings to see from where they are referenced |
| 19:59 | <zewt> | annevk: btw, something still broken in dom4 |
| 20:00 | <zewt> | not sure if you repro'd that |
| 20:00 | <zewt> | (with xrefs, I mean, not just "something") |
| 20:00 | <zewt> | everyone's favorite bug report. "something is broken" |
| 20:00 | <annevk> | they work for me |
| 20:01 | <zewt> | doesn't work for me in FF8 or Chrome |
| 20:01 | <annevk> | works in Chrome too |
| 20:01 | <zewt> | https://zewt.org/~glenn/temp.png |
| 20:01 | <annevk> | also Gecko |
| 20:02 | <annevk> | so your problem is the CSS? |
| 20:02 | <zewt> | missing css? |
| 20:02 | <annevk> | well yes |
| 20:02 | <annevk> | but that doesn't seem so important |
| 20:02 | <annevk> | feel free to figure out a patch if you care |
| 20:02 | <zewt> | huh? it's dumping text in the middle of a line, instead of doing what the other specs do |
| 20:03 | <annevk> | it still works |
| 20:04 | <zewt> | is this stuff reinvented for each spec? heh |
| 20:06 | <annevk> | zewt: it just doesn't use the WHATWG style sheet |
| 20:06 | <annevk> | zewt: and since I haven't really found it a problem, I haven't tried figuring out the details |
| 20:07 | <annevk> | MikeSmith: https://www.w3.org/Bugs/Public/show_bug.cgi?id=10694 should be moved to the browser testing group but I cannot find that :) |
| 20:07 | <MikeSmith> | annevk: OK, will fix that |
| 20:07 | <annevk> | MikeSmith: and https://www.w3.org/Bugs/Public/show_bug.cgi?id=14569 needs a URL component somewhere |
| 20:07 | <annevk> | but that can wait |
| 20:08 | <MikeSmith> | k |
| 20:12 | <jamesr> | somewhat dumb spec question: is the list here http://www.whatwg.org/specs/web-apps/current-work/multipage/elements.html#interactive-content-0 supposed to be an example or exhaustive? |
| 20:12 | <jamesr> | i'm pretty sure that's what you run through when clicking on a <div> with an onclick handler, but if that list is supposed to be exhaustive i'm not sure if that div counts as 'interactive content' |
| 20:12 | <annevk> | exhaustive |
| 20:13 | <jamesr> | so what blesses the div with activation behavior? |
| 20:13 | <jamesr> | registering the onclick handler in DOM? |
| 20:14 | <annevk> | nothing I think |
| 20:14 | <annevk> | oh sorry |
| 20:14 | <jamesr> | ok, so if it doesn't have activation behavior then according to that part of the text click events shouldn't fire on it |
| 20:15 | <jamesr> | or does the click event propagation still happen, just not via this algorithm? |
| 20:15 | <annevk> | it does have activation behavior, but which elements have activation behavior the spec does not define |
| 20:15 | <jamesr> | ok, so someone somewhere else (presumably dom events) is supposed to say that the element has activation behavior |
| 20:15 | <jamesr> | when such a handler is registered |
| 20:15 | <jamesr> | right? |
| 20:16 | <annevk> | dunno really what the exact plan for that is |
| 20:17 | <annevk> | but yeah ideally some spec would define that |
| 20:17 | <annevk> | because currently it's a big guessing game when implementing a UA for mobile phone and you lack a mouse... |
| 20:17 | <annevk> | or if you want to make a site keyboard accessible |
| 20:18 | <annevk> | Opera's spatial navigation is a pretty good attempt at that I think and I tried documenting the heuristics once, but it was complicated |
| 20:20 | <smaug____> | jamesr: btw, I think activation behavior is largely non-reviewed in the spec, since it isn't implemented in any browser |
| 20:21 | <smaug____> | (unless it has been implemented in some browser during last month or so) |
| 20:53 | <annevk> | re Chrome and ISO-2022-KR, it seems they supported windows-949 already and just use that? |
| 21:01 | <annevk> | http://codereview.chromium.org/6010003/diff/6001/icu46/source/data/mappings/noop-cns-11643.ucm is interesting |
| 21:02 | <annevk> | http://codereview.chromium.org/6010003/diff/6001/icu46/source/data/mappings/noop-iso-ir-165.ucm |
| 21:02 | <annevk> | there's a couple of those |
| 21:02 | <annevk> | I guess they are for the issue Philip` mentioned earlier |
| 21:06 | <annevk> | MikeSmith: would having a "WHATWG" product go too far? so we can have Encoding / URL / Fullscreen as components there for now? |
| 21:06 | <annevk> | MikeSmith: or maybe WGLESS if we want something neutral |
| 21:08 | <kennyluck> | or COMMUNITY |
| 21:09 | <zewt> | *COUGH*WG |
| 21:09 | <annevk> | Community works, or "Platform" :) |
| 22:16 | <jamesr> | TabAtkins: you've lied, svg is at least slightly retarded |
| 22:16 | <hober> | annevk: re: https://www.w3.org/Bugs/Public/show_bug.cgi?id=15213 per the layering argument you make there, would you say that https://bugs.webkit.org/show_bug.cgi?id=73237 should be WONTFIXed? |
| 22:16 | <TabAtkins_> | I didn't say it wasn't completely retarded. Just that doing basic diagrams with it shouldn't be troublesome cross-browser. |
| 22:17 | <shepazu> | SVG is not retarded, it's just… special. |
| 22:17 | <jamesr> | ok, so i have a simple diagram with some <rect>s |
| 22:17 | <heycam> | TabAtkins_, needs more layout! |
| 22:18 | <jamesr> | i'm using colors (fill) to represent different types of things |
| 22:20 | <annevk> | hober: yes, that's exactly the same problem |
| 22:20 | <annevk> | hober: the point of ARIA was a low-level accessibility API |
| 22:20 | <annevk> | hober: not one which has semantics tied to it |
| 22:20 | <annevk> | is* |
| 22:22 | <jamesr> | for the curious, tab solved my problem offline. answer = use CSS for indirection, don't use SVG's indirection mechanisms |
| 22:23 | <shepazu> | jamesr: what was the question? too many answers, not enough questions... |
| 22:24 | <zewt> | svg jeopardy |
| 22:24 | <TabAtkins_> | james didn't understand the dependency graph that SVG uses. |
| 22:24 | <jamesr> | ah, didn't type it out. question was i want a group of <rects> that are style the same way (solid color), but i want an easy way to toggle them all to a different type of display (pattern) |
| 22:25 | <jamesr> | so i can have a color and color-blind-friendly diagram with the same SVG DOM |
| 22:25 | <heycam> | jamesr, CSS sounds like the right answer there! |
| 22:25 | <shepazu> | yup |
| 22:25 | <roc> | I'm not sure why you would try anything other than CSS for that |
| 22:25 | <TabAtkins_> | ^_^ |
| 22:26 | <jamesr> | well, i didn't know you could reference SVG patterns, etc, from CSS |
| 22:26 | <roc> | ah |
| 22:28 | <hober> | annevk: *nod* |
| 22:30 | <TabAtkins_> | heycam, annevk: in the webrtc list, I complained about the use of numeric constants, and asked them to change to strings. They've pushed back, and are asking for the arguments behind the change in webidl, and any official resolution. |
| 22:30 | <TabAtkins_> | I'm not sure what to search for to find them myself. |
| 22:30 | <hober> | annevk: thanks |
| 22:30 | <zewt> | TabAtkins: always nice to have to repeat every. single. decision., every single time |
| 22:31 | <TabAtkins_> | zewt: Indeed. |
| 22:31 | <zewt> | seems like a corollary of NIH |
| 22:32 | <heycam> | TabAtkins_, numeric constants don't buy you anything, they're longer to type (assuming you use the constants and not just "0"!), they require properties to exist (where strings do not) |
| 22:33 | <heycam> | TabAtkins_, they are just as self descriptive |
| 22:33 | <zewt> | they needlessly create two ways to do everything (by naming the constant, or by hardcoding the property) |
| 22:33 | <zewt> | (value) |
| 22:33 | <zewt> | eg. creating a "bad" way to do things in addition to a "good" way |
| 22:33 | <TabAtkins_> | Why is requiring properties to exist a bad thing? Just more memory to maintain? |
| 22:34 | <annevk> | the bad thing is that people will use numbers instead |
| 22:34 | <heycam> | TabAtkins_, yeah, it's a very slight negative |
| 22:34 | <annevk> | and constants are more verbose in actual usage |
| 22:34 | <TabAtkins_> | heycam: Can you find the email where you resolved on it, too? |
| 22:35 | <heycam> | TabAtkins_, there was no official resolution, since it's more of a style/design issue |
| 22:35 | <zewt> | annevk: yeah, creating a "bad" way that's less typing than the "good"--which is an incentive for inexperienced programmers |
| 22:35 | <zewt> | (to choose the bad way) |
| 22:35 | <TabAtkins_> | heycam: Ok. |
| 22:35 | <heycam> | TabAtkins_, there's also: http://scriptlib-cg.github.com/api-design-cookbook/#don-t-use-numerical-constants |
| 22:36 | <zewt> | named symbols that don't have a visible integer value wouldn't be as evil, but we don't have those |
| 22:36 | <TabAtkins_> | zewt: We do, but they're called "strings". |
| 22:36 | <zewt> | those aren't symbols |
| 22:37 | <TabAtkins_> | Close enough. ^_^ |
| 22:37 | <annevk> | number (very bad); constants (verbose); strings (perfect) |
| 22:37 | <zewt> | referring to eg. "feature = object()" in Python |
| 22:37 | <annevk> | it's not very hard |
| 22:37 | <TabAtkins_> | And they're likely the same as symbols underneath - atomic strings or something. |
| 22:37 | <annevk> | yeah you can intern them |
| 22:37 | <zewt> | and i'd expect that they'd be interned anyway, since they're constants in the code |
| 22:37 | <annevk> | and once heycam adds string enum you can even have them natively in IDL |
| 22:38 | <zewt> | (lua does that, I think Python does too) |
| 22:38 | <zewt> | (well, lua interns all strings, so I guess that's not interesting) |
| 22:38 | <TabAtkins_> | Yeah, I don't know how modern JS engines intern string constants in code. |
| 22:40 | <zewt> | i don't know how interning works in JS anyway, since it's transparent in JS (as far as I know?), where you can tell the difference if you want to in Python (is) |
| 22:40 | <TabAtkins_> | Yeah. |
| 22:41 | <zewt> | (eg. it's more useful to know the low-level details in Python than JS) |
| 22:41 | <jamesr> | in JS strings are value types, not object types |
| 22:41 | <jamesr> | so you can't tell |
| 22:42 | <sythe> | Hey |
| 22:42 | <sythe> | <audio loop autoplay> <source src="up.wav"> </audio> |
| 22:42 | <sythe> | I'm using that to play an audio file |
| 22:42 | <sythe> | Why doesn't it loop? |
| 22:47 | <TabAtkins_> | It should. Example page? |
| 22:59 | <sythe> | TabAtkins, http://openzest.com/s/game/game3.html |
| 23:21 | <TabAtkins_> | sythe: Works just fine for me, as far as I can tell. |
| 23:21 | <sythe> | TabAtkins, Ah, yes |
| 23:21 | <sythe> | It doesn't, TabAtkins |
| 23:21 | <sythe> | I cheated, and used a 10 minute sound file |
| 23:21 | <TabAtkins_> | I know. |
| 23:21 | <TabAtkins_> | When I opened one of the files myself and added loop, it looped correctly. |
| 23:21 | <sythe> | Just try it in a real-time html editor |
| 23:21 | <sythe> | Really? |
| 23:21 | <sythe> | It doesn't work at all |
| 23:22 | <TabAtkins_> | Yes. Chrome tries to open .ogg in <video> rather than <audio>, but still. |
| 23:22 | <sythe> | In Firefox and/or Chrome/Safari |
| 23:22 | <smaug____> | TabAtkins_: hey, when will Chrome finally drop h264 ? |
| 23:23 | <TabAtkins_> | sythe: I just set up a proper <audio> pointing to the mp3, and it also worked fine in Chrome. |
| 23:23 | <TabAtkins_> | smaug____: No comment. We're talking with roc about details. |
| 23:24 | <sythe> | TabAtkins, With autoplay? |
| 23:24 | <smaug____> | ah, ok, understood |
| 23:25 | <TabAtkins_> | sythe: Yes. |
| 23:25 | <TabAtkins_> | <audio controls autoplay loop src="whatever-your-mp3-file-was"></audio> |
| 23:26 | <sythe> | Well |
| 23:26 | <sythe> | It doesn't work in Safari, does it? |
| 23:26 | <TabAtkins_> | http://software.hixie.ch/utilities/js/live-dom-viewer/saved/1286 |
| 23:27 | <TabAtkins_> | No idea. I don't have Safari on hand. |
| 23:27 | <sythe> | TabAtkins, Is the "preload" attribute messing it up |
| 23:27 | <sythe> | ? |
| 23:27 | <sythe> | I've tried it with Chrome, Safari and FF |
| 23:27 | <sythe> | None of them loop it |
| 23:27 | <sythe> | (On Linux, BTW) |
| 23:28 | <roc> | we only added "loop" support in FF11 |
| 23:28 | <smaug____> | and we don't support mp3 |
| 23:29 | <roc> | that too |
| 23:29 | <sythe> | roc, Really? |
| 23:29 | <TabAtkins_> | sythe: preload doesn't affect anything on my computer. I'm running Chrome 16 on Linux. |
| 23:30 | <roc> | yes |
| 23:30 | <sythe> | TabAtkins, Does it loop? In Chrome 16 on Linux? |
| 23:30 | <TabAtkins_> | sythe: Yes. |
| 23:30 | <sythe> | Lemme double-check |
| 23:31 | <sythe> | It doesn't even play for me, actually |
| 23:31 | <sythe> | In Chromium |
| 23:31 | <TabAtkins_> | sythe: then you're doing something exceedingly wrong, I guess. ^_^ |
| 23:31 | <sythe> | TabAtkins, I had other testers |
| 23:31 | <sythe> | The only person who had it looping was on OSX |
| 23:31 | <sythe> | And he said that it made a beeping noise between loops |
| 23:32 | <TabAtkins_> | All I know is that duplicating the element in live dom viewer and skipping to the end makes it loop. And it played in the original page (though I don't think I was on it long enough to hear if it looped). |
| 23:32 | <TabAtkins_> | There was a small skip between loops, but no beep for me to hear. |
| 23:32 | <TabAtkins_> | (That skip is possibly a bug - it should be seamless if the sounds are.) |
| 23:35 | <sythe> | Hmm |
| 23:35 | <sythe> | So...will it loop in Firefox? |
| 23:36 | <sythe> | "we only added "loop" support in FF11...and we don't support mp3" |
| 23:36 | <sythe> | That's a bit of a fail |
| 23:37 | <sythe> | If true, it basically makes it unable for me to loop audio, or perhaps even use HTML5 audio, at all, ATM |
| 23:39 | <sythe> | Right? |
| 23:39 | <TabAtkins_> | Nah, you can use script to catch when the audio ends, and start it again. |
| 23:39 | <TabAtkins_> | Not supporting mp3 is actually rather reasonable, given the licensing issues. |
| 23:40 | <sythe> | TabAtkins, Link...to this fabled script? |
| 23:40 | <TabAtkins_> | Restarting it with script isn't as good as the "loop" attribute (you're guaranteed a small skip), but it'll work. |
| 23:40 | <sythe> | None of the ones I tried seemed to work |
| 23:41 | <TabAtkins_> | sythe: A trivial one would be to create a setInterval with the same delay as the length of the audio that resets its time to 0. |
| 23:43 | <jamesr> | is there an event fired when the playing ends? |
| 23:43 | <TabAtkins_> | I think so, but I can't recall its name and can't find it in the spec right now. |
