| 07:04 | <hsivonen> | TabAtkins_: I think it's insulting that they keep saying that anyone can implement clearkey as if Boris and I were dumb enough to accept that as addressing our concerns. |
| 07:19 | <Hixie> | clearkey is a red herring. the spec has a better solution to the problem clearkey solves. |
| 07:45 | <charlvn> | does anyone have a url for clearkey? googled but found a bunch of other stuff that doesn't seem to be related to tech or standards |
| 08:38 | <hsivonen> | charlvn: http://dvcs.w3.org/hg/html-media/raw-file/tip/encrypted-media/encrypted-media.html#simple-decryption-clear-key |
| 08:40 | <hsivonen> | great. ISO/IEC 23001-7:2012 is not available for HTTP GET without 74 CHF |
| 08:41 | <hsivonen> | does an mkv equivalent of ISO Common Encryption exist? |
| 08:41 | <hsivonen> | it seems Matroska does have encryption hooks for DRM |
| 08:44 | <MikeSmith> | hsivonen: found http://www.3gpp.org/ftp/Inbox/LSs_from_external_bodies/ISO_IEC_JTC1_SG29_WG11/29n12316.zip |
| 08:46 | <hsivonen> | MikeSmith: thanks |
| 08:46 | <hsivonen> | I wonder if that's an RF spec |
| 08:47 | <hsivonen> | what's the main benefit of putting AES-CTR inside the container as opposed to encrypting the whole file using AES-CTR? |
| 09:25 | <hsivonen> | I'm rather annoyed that it took this long in the thread for the requirement of integrating with ISO Common Encryption to show up |
| 10:07 | <charlvn> | hsivonen: thanks! |
| 10:45 | <annevk> | reading is hard |
| 10:45 | <annevk> | this time it's not me, but some guy on ietf-http-wg though |
| 12:28 | <matjas> | was `document.scripts` specced before “HTML5”? http://www.whatwg.org/specs/web-apps/current-work/multipage/dom.html#dom-document-scripts |
| 12:30 | <hsivonen> | matjas: if it had been, it would probably have a more complicated name than "scripts" |
| 12:30 | <matjas> | hah |
| 12:50 | <hsivonen> | hmm. no new champions in http://www.ie6countdown.com/champions.aspx in January and February |
| 13:03 | <annevk> | https://twitter.com/molily/status/176652243901222912 |
| 13:04 | <annevk> | What else is somewhat bad? |
| 13:30 | <zcorpan> | http://www.whatwg.org/specs/web-apps/current-work/multipage/semantics.html#attr-link-href - why is this non-empty URL? <a> allows empty |
| 13:36 | <hsivonen> | Do I interpret this right: https://twitter.com/#!/johnfoliot/status/176323582195023872 JF doesn't want users to have the right to run a speech recognizer on Netflix audio, because supporting such right would acknowledge that Netflix might not always provide human-written captions |
| 13:37 | <hsivonen> | and JF wants to insist that Netflix must provide human-written captions |
| 13:46 | <jgraham> | hsivonen: Do you feel you get value out of arguing with JF? |
| 14:46 | <woef> | Does anyone know why Safari has a non-conform UA string? |
| 14:47 | <woef> | Other browsers have "Browser/versionnr" |
| 14:47 | <woef> | While Safari has "Version/versionnr Safari/webkitversion" |
| 14:51 | <hsivonen> | woef: probably because they first put a non-marketing version after Safari/ and by the time they wanted to put a marketing version in the UA string, too, sites already depended on the non-marketing version after Safari/ |
| 14:53 | <woef> | Annoying though, makes it hard to get decent stats from GA. |
| 14:53 | <woef> | Guess they won't change it anytime soon then. |
| 14:53 | <Philip`> | Why can't GA just parse the UA string in an appropriate way? |
| 14:53 | <woef> | Make an exception for Safari? |
| 14:54 | <woef> | You'll have to ask Google I guess :) |
| 14:54 | <woef> | Other browsers all work in the same way. |
| 14:54 | <woef> | Browser/version nr |
| 14:55 | <Philip`> | Opera doesn't |
| 14:56 | <Philip`> | Any attempt to treat UA strings as anything other than opaque string templates with occasional optional/version-dependent tokens seems doomed to failure |
| 15:48 | <zcorpan> | http://dev.w3.org/html5/html4-differences/Overview.src.html has lots of new text, review much appreciated :-) |
| 15:49 | <zcorpan> | commit log http://dev.w3.org/cvsweb/html5/html4-differences/Overview.src.html |
| 15:49 | <annevk> | oh shall I generate a copy for you? |
| 15:50 | <zcorpan> | sure |
| 15:50 | <zcorpan> | i might take a stab at https://www.w3.org/Bugs/Public/show_bug.cgi?id=13463 tonight also |
| 15:50 | <annevk> | hmm |
| 15:50 | <annevk> | isindex not defined |
| 15:51 | <zcorpan> | http://www.whatwg.org/specs/web-apps/current-work/multipage/tree-construction.html#isindex ? |
| 15:52 | <zcorpan> | oh |
| 15:52 | <zcorpan> | it should point to http://www.whatwg.org/specs/web-apps/current-work/multipage/obsolete.html#isindex-0 no? |
| 15:53 | <annevk> | the term database is missing lots of terms |
| 15:54 | <annevk> | this might be a nice distraction from studying gb18030 though |
| 15:59 | <annevk> | ms2ger: feature request, find out all the undefined terms in a document rather than just the first |
| 16:01 | <bga> | annevk is opera userjs events available in nonuserjs? |
| 16:02 | <bga> | im interesting about onBeforeScropt/onAfterScript |
| 16:02 | <jgraham> | No. Unless there is a bug. |
| 16:03 | <rahaman> | Hi all...i need one clarification reg how to specify the border attributes for <td> element, some time back I had sent an email on the same.. |
| 16:03 | <rahaman> | can any one please help me out ? |
| 16:06 | <annevk> | bga: what do you want them for? |
| 16:06 | <annevk> | bga: we were thinking of adding the events in HTML, but the use cases seem somewhat unclear |
| 16:07 | <bga> | i see |
| 16:17 | <annevk> | where did zcorpan go? |
| 16:35 | <Hixie> | annevk: just do a cg for the encoding thing |
| 16:36 | <annevk> | my employer prefers WGs |
| 16:53 | <Hixie> | hsivonen: i don't see a requirement anywhere to encrypt the frames rather than the entire stream |
| 16:54 | <Hixie> | hsivonen: the only requirement that was raised similar to that was that some formats use a manifest that itself points to the streams on the CDN that _are_ encrypted, but that's handled by this proposal -- you just put the http+aes urls in the manifest instead of http: urls |
| 17:19 | <AryehGregor> | Yay, the CSSWG switched to hg just in time. |
| 17:19 | <AryehGregor> | Looks like I won't have to learn CVS after all. |
| 17:44 | <Ms2ger> | MikeSmith: |
| 17:44 | <Ms2ger> | remote: importing Hg objects into Git |
| 17:44 | <Ms2ger> | remote: abort: Permission denied: /var/hg/repos/html/.git/objects/9a/cdb4fbb68fecbb5b9d7bd45c4f98e65f905f75.lock |
| 17:44 | <Ms2ger> | remote: fatal: Unable to create '/var/hg/repos/html/.git/refs/remotes/github/master.lock': Permission denied |
| 17:45 | <Ms2ger> | remote: warning: changegroup.github hook exited with status 128 |
| 17:56 | <dglazkov> | good morning, Whatwg! |
| 18:20 | <abarth> | Hixie: looking at http://dev.w3.org/html5/spec/Overview.html#http-aes-scheme . I'm wondering why there isn't a use case for integrity as well |
| 18:21 | <abarth> | Hixie: e.g., hosting a script on an untrusted CDN |
| 18:21 | <Hixie> | because if the CDN were to screw around and change things, they'd be sued and lose the business |
| 18:22 | <Hixie> | whereas just stealing information can be done without leaving a trace |
| 18:22 | <abarth> | as an example, we could not show a mixed content indicator if you loaded an integrity-checked script from http+aes on an HTTPS page |
| 18:22 | <abarth> | s/could not/would not have to/ |
| 18:23 | <Hixie> | the headers would still be mitm'able |
| 18:23 | <Hixie> | so i don't think that's true |
| 18:23 | <abarth> | today we show a mixed content warning because we're worried about an active network attacker corrupting the contents of the script and XSSing the page |
| 18:23 | <abarth> | i guess there's some trickiness with redirects |
| 18:24 | <abarth> | we'd need to make sure the final resource passed the integrity check |
| 18:24 | <Hixie> | as well as all those in between |
| 18:25 | <Hixie> | since an integrity check is easy to fake if you're in charge of the resource being served |
| 18:25 | <abarth> | do you mean the URL of the resource or the resource itself? |
| 18:25 | <Hixie> | both or either, i guess |
| 18:25 | <Hixie> | i think if integrity matters, you should use https+aes |
| 18:26 | <Hixie> | why reinvent the wheel |
| 18:26 | <abarth> | yeah, the only reason is performance |
| 18:26 | <abarth> | but HTTPS probably has fine performance now-a-days |
| 18:26 | <Hixie> | given that google.com is served on https now, i think the performance thing can be put to bed now |
| 18:30 | <abarth> | ekr: the question is mainly what the use case is for integrity checking |
| 18:31 | <abarth> | ekr: we've been talking about the mixed content use case, but that seems better addressed by HTTPS |
| 18:32 | <ekr> | Let me ask the question differently. What's the argument *against* integrity checking. |
| 18:32 | <Ms2ger> | Hehe |
| 18:33 | <Ms2ger> | The burden of proof lies with the person arguing for *more* features |
| 18:33 | <abarth> | ekr: i would think there are two reasons: |
| 18:33 | <abarth> | 1) it's more complexity |
| 18:33 | <ekr> | While that may be generally true, there's ample experience that lack of integrity checking has led to serious security problems in a number of protocols |
| 18:33 | <abarth> | 2) redirects and headers can already screw with integrity, so we shouldn't try to promising properties we can't deliver |
| 18:34 | <ekr> | Most recently, for instance, look at the security problems with XML encryption. |
| 18:35 | <ekr> | So since integrity is the common security idiom, I think some argument is required for why it should not be used. |
| 18:36 | <zcorpan> | annevk: i went to fix food |
| 18:36 | <ekr> | WRT to the points you raise, Adam, I'm not sure I understand the setting here. I.e., if I have an http+aes URI, and I get redirected, does that mean that I expect the new target not to be encrypted? |
| 18:39 | <abarth> | ekr: correct |
| 18:40 | <abarth> | (or at least that's what I've gathered from reading what in the spec today) |
| 18:41 | <ekr> | Without taking a position on whether that is a good design, integrity is about more than total message forgery. So, without integrity protection an attacker could (for instance) modify specific portions of a message while leaving some values intact. This is particularly easy with CTR, where it's possible to flip any bit. |
| 18:42 | <ekr> | I'm not saying that I have a concrete attack, but since any data can be carried over this channel, it's easy to see that an active attacker might be able to make significant semantic changes. |
| 18:42 | <ekr> | Just because you don't know the entire content of the message doesn't mean you don't know parts of it. |
| 18:43 | <ekr> | But I think more importantly, as the XMLENC case shows, lack of integrity can be a threat to confidentiality |
| 18:45 | <Ms2ger> | annevk, fixed |
| 18:48 | <AryehGregor> | Is hixie.ch down? |
| 18:54 | <zcorpan> | whatwg.org seems down for me |
| 18:55 | <charlvn> | http://www.downforeveryoneorjustme.com/whatwg.org |
| 18:55 | <charlvn> | seems to be up for me |
| 18:56 | <divya> | everything is down because dreamhost is down |
| 18:57 | <AryehGregor> | Back up for me. |
| 18:58 | <divya> | yep all back up seems like. |
| 19:08 | <charlvn> | down again |
| 19:29 | <annevk> | Ms2ger: oooh you fixed that bug? I might actually complete html5-diff for zcorpan then |
| 19:29 | <annevk> | html5.org is up |
| 19:30 | <annevk> | Hixie: processing model of http+aes should probably be somewhat more clearly defined; people get confused about it |
| 19:50 | <annevk> | arv: big fan of the non-existing method(...magic) huh? :) |
| 19:50 | <arv> | annevk: Yup |
| 19:50 | <annevk> | arv: I guess we can make it vararg, but the += argument doesn't work for remove() and would not quite do the same thing (duplication) |
| 19:51 | <annevk> | arv: and without ...magic we'd be kind of lost, but maybe that makes the case for postponing work on this feature some more :) |
| 19:51 | <arv> | annevk: Duplication does not really matter |
| 19:52 | <arv> | annevk: it is one more reason why js libs are faster |
| 19:52 | <annevk> | are they really faster? |
| 19:52 | <annevk> | it sounds kind of silly |
| 19:52 | <arv> | yes |
| 19:52 | <arv> | but like I said in the bug, the js libs do not validate the input |
| 19:52 | <annevk> | are the implementations not optimized or something or just optimized for string manipulation? |
| 19:53 | <zcorpan> | annevk: yo |
| 19:53 | <arv> | the webkit one is faster for remove and has because we keep a vector of AtomicStrings (pointer comparison) |
| 19:55 | <miketaylr> | i think http://jsperf.com/classlist-vs-addclass shows that classList stuff can be faster, depending on the browser |
| 19:56 | <miketaylr> | i guess faster in safari and opera, slower in ff and chrome |
| 19:57 | <arv> | miketaylr: Yeah, Chrome DOM bindings are slow so doing a single className assignment is much faster than multiple adds |
| 19:59 | <annevk> | yay Opera |
| 20:01 | <annevk> | zcorpan: are you still updating? |
| 20:02 | <zcorpan> | annevk: working on https://www.w3.org/Bugs/Public/show_bug.cgi?id=13463 |
| 20:02 | <zcorpan> | haven't committed yet |
| 20:06 | <annevk> | Ms2ger: I get "SyntaxError: Specification not found: ." on html5-diff now |
| 20:06 | <annevk> | Ms2ger: after pulling, updating and running setup.py install |
| 20:07 | <annevk> | and afaict nothing is wrong with the data-anolis-ref links |
| 20:13 | <annevk> | ooh |
| 20:13 | <annevk> | wait |
| 20:13 | <annevk> | that is about data-anolis-spec being wrong |
| 20:16 | <annevk> | not a bug in Anolis |
| 20:16 | <annevk> | but now I fixed that in Overview.src.html and there's a huge list of undefined terms :( |
| 20:19 | <Hixie> | my plan to not read e-mail today and _just_ work on canvas is going to be apparently significantly helped today by dreamhost |
| 20:19 | <annevk> | email is down? |
| 20:20 | <Hixie> | apparently my mail server isn't responding, yeah |
| 20:20 | <Hixie> | dunno why |
| 20:20 | <Hixie> | it was working earlier |
| 20:20 | <Hixie> | doesn't matter though, i really _was_ planning to ignore my e-mail :-) |
| 20:20 | <AryehGregor> | Does anyone have any ideas on how to automatically test performance of transitions/animations cross-browser? My suspicion is it's not possible. |
| 20:20 | <AryehGregor> | At least, not usefully. |
| 20:20 | <zcorpan> | annevk: what needs changing in src.html? |
| 20:20 | AryehGregor | looks at requestAnimationFrame |
| 20:21 | <annevk> | zcorpan: search for data-anolis-spec> |
| 20:21 | <Hixie> | AryehGregor: what are the requirements you are trying to test? |
| 20:21 | <annevk> | zcorpan: change it into data-anolis-spec=html> |
| 20:21 | <zcorpan> | annevk: thanks, fixed |
| 20:21 | <zcorpan> | (locally) |
| 20:21 | <Hixie> | AryehGregor: and do you mean in a self-contained automated fashion, manually, or with external automation help? |
| 20:21 | <AryehGregor> | Hixie, I don't think there are any spec requirements here, it's more like a QoI test. Like how Acid3 said it had to animate smoothly, but more automatically/objectively than that. |
| 20:22 | <AryehGregor> | TBH I'm not sure what the requirements are, exactly. |
| 20:22 | <AryehGregor> | Maybe I should figure them out. |
| 20:22 | <Hixie> | AryehGregor: it's usually a good start to writing a test :-) |
| 20:22 | <jgraham> | AryehGregor: If you do, let me know :) It seems like an intersting thing to test even in a browser-specific fashion |
| 20:22 | <zcorpan> | seems opera still supporst HTMLIsIndexElement |
| 20:23 | <jgraham> | But I don't think we have (m)any QoI tests like that for Opera |
| 20:23 | <AryehGregor> | The browser probably knows how often it's doing repaints or such, and should be able to secretly figure out if it's more or less than some threshold. |
| 20:23 | <jgraham> | I would like us to of cousre |
| 20:23 | <AryehGregor> | But not in a browser-independent fashion. I don't think there's ever going to be any reliable way to test FPS other than just looking. |
| 20:23 | <AryehGregor> | Possibly the looking could be automated by taking screenshots automatically or such, of course. |
| 20:23 | <Hixie> | AryehGregor: if you can use external tools, it seems you might be able to automate it by having the test set up to flash specific patterns at specific intervals, and then record it using an external video recorder (physical or screen grab tool) and having a script analyse the video to make sure the patterns appear at the right time |
| 20:23 | <Hixie> | AryehGregor: but that may be a bit over the top in complexity |
| 20:24 | <jgraham> | Relying on a physical camera seems like a non-starter |
| 20:24 | <AryehGregor> | Or just estimate FPS by taking a screenshot every 10ms or something and checking when two consecutive screenshots match. |
| 20:24 | <jgraham> | Well maybe not if you are Google |
| 20:24 | <jgraham> | But taking a screenshot consumes resources itself |
| 20:24 | <Hixie> | AryehGregor: someone's working on a set of testing apis, that might be a place to start to add some sort of way to report the frame rate or something |
| 20:24 | <AryehGregor> | Of course, that assumes browsers aren't doing weird small things that make screenshots not match. |
| 20:25 | <AryehGregor> | What incentive do browsers have to ever claim less than 60fps? |
| 20:25 | <AryehGregor> | Or whatever is considered "high". |
| 20:25 | <jgraham> | Hixie: You mean the WebDriver stuff? |
| 20:25 | <Hixie> | AryehGregor: if it's a testing api, that only ships in test builds or is only enabled in test mode, it seems useful for them to be honest |
| 20:25 | <Hixie> | jgraham: yeah |
| 20:25 | <jgraham> | I don't think that helps here except that it provides an API for taking screenshots |
| 20:25 | <AryehGregor> | Not if it makes them look bad relative to other browsers. |
| 20:25 | <AryehGregor> | Oh, it does? Privileged, I guess? |
| 20:25 | <Hixie> | AryehGregor: (with my idea the patterns could be qr codes that are quite resilient to rendering errors, if you just wanted to test timing) |
| 20:26 | <jgraham> | Tah's not really how it works |
| 20:26 | <jgraham> | WebDriver, I mean |
| 20:26 | <jgraham> | It is mostly a browser-control API |
| 20:26 | <Hixie> | jgraham: i meant that he could add new features to webdriver if necessary |
| 20:26 | <jgraham> | Sure, but it seems quite different to the existing feature set |
| 20:26 | <jgraham> | and the existing feature set generally works in nonmal builds |
| 20:26 | <jgraham> | *normal |
| 20:27 | <jgraham> | At least in opera |
| 20:27 | <gsnedders> | There's no guarantee as to how quickly a screenshot will be taken normally. |
| 20:27 | <Philip`> | AryehGregor: Artifical microbenchmarks sound generally bad because you'll encourage developers to optimise for cases that won't matter to real users, so it'd be better to construct a small number of full real-world demos that can be run manually to see if there's user-perceivable performance problems |
| 20:27 | <gsnedders> | So using it for timing dependant stuff is questionable. |
| 20:27 | <Philip`> | and then you wouldn't need an automatic mechanism for it |
| 20:27 | <Hixie> | yeah if you want to do it manually it's a lot easier |
| 20:27 | <Hixie> | (that's what acid3 did, basically) |
| 20:28 | <Hixie> | (not an especially good part of acid3, but that's another story) |
| 20:28 | <jgraham> | Right, I am a bit skeptical of a screenshot-based approach |
| 20:28 | <AryehGregor> | Philip`, kind of like IE Test Drive except not subtly biased to perform as badly as possible in all other browsers? |
| 20:28 | <Philip`> | AryehGregor: Yes |
| 20:29 | AryehGregor | generously assumes that they wrote the tests without deliberate bias and then just optimized IE as much as possible for them before publicizing them, rather than deliberately sabotaging other browsers' performance |
| 20:29 | <jgraham> | AryehGregor: I thought they all tested the same thing (bliting perf.) |
| 20:29 | <AryehGregor> | Also possible. |
| 20:29 | <AryehGregor> | I didn't look at the details. |
| 20:29 | <jgraham> | Which just so happened to be the one thing that IE was good at compared to the others |
| 20:29 | <jgraham> | I thought one of our gfx guys said that, but I might be misremembering |
| 20:30 | Philip` | remembers hearing the same comment from somewhere |
| 20:30 | <gsnedders> | jgraham: That is true |
| 20:31 | <Philip`> | If blitting performance is what real web pages will care about, then that's a useful thing to test, but I presume it's not the bottleneck in every likely case |
| 20:32 | <annevk> | zcorpan: also it's attr-hyperlink-usemap |
| 20:33 | <jgraham> | Anyway, it seems like you could do something like take a screenshot regulsr, but quite large, intervals and measure expected_position - actual_position in each screenshot, then subtract off some constant component (to allow for the time taken to create the screenshot) |
| 20:33 | <annevk> | zcorpan: and dom-document-onreadystatechange should be handler-onreadystatechange |
| 20:34 | <jgraham> | Seems pretty hard, but not impossible to do. I don't know if it would give good results though |
| 20:36 | <jgraham> | I guess one issue is that to make it work, you would need to make the animation very simple, which would probably also make it performant |
| 20:36 | <annevk> | zcorpan: commandtype is dom-command-ro-commandtype |
| 20:38 | <annevk> | zcorpan: likewise for the other commands |
| 20:45 | <annevk> | zcorpan: also not attr-form-enctype but attr-fs-enctype |
| 20:46 | <annevk> | zcorpan: is async supposed to be attr-script-async? |
| 20:47 | <annevk> | zcorpan: ilnk -> link |
| 20:48 | <annevk> | zcorpan: attr-body-background -> attr-background |
| 20:48 | <zcorpan> | annevk: thanks! keep em coming, i'll read the logs. bbiab |
| 20:49 | <annevk> | zcorpan: form-method -> fs-method |
| 20:50 | <annevk> | zcorpan: what is concept-categories? |
| 20:50 | <annevk> | form-action -> fs-action |
| 20:51 | <annevk> | zcorpan: you have used both 'dir' and 'attr-dir' |
| 20:52 | <annevk> | attr is correct |
| 20:52 | <annevk> | attr-hyperlink-name -> attr-a-name |
| 20:56 | <annevk> | zcorpan: I updated specification-data now; once you fix the mistakes I can probably finish this |
| 21:05 | <MikeSmith> | Ms2ger: thanks, will fix that now |
| 21:08 | <MikeSmith> | Ms2ger: fixed |
| 21:11 | <jgraham> | MikeSmith: Thanks |
| 21:11 | <jgraham> | Even though I am not Ms2ger |
| 21:11 | <MikeSmith> | heh |
| 21:12 | <MikeSmith> | I think it should all work again now |
| 21:29 | <annevk> | @WHATWG has 6000 followers |
| 21:41 | <Ms2ger> | MikeSmith, thanks, though I don't care about git ;) |
| 21:42 | <Ms2ger> | annevk, no bugs on my side for Anolis? |
| 21:42 | <annevk> | nope |
| 21:45 | <zcorpan> | annevk: thanks |
| 21:49 | <jgraham> | Ms2ger: You might care about the exisance of code-review on github |
| 21:49 | <jgraham> | *existence |
| 21:50 | <Ms2ger> | That's a nice typo |
| 22:00 | zcorpan | realizes there's no idl attribute that reflects form="" |
| 22:00 | <TabAtkins_> | From @mattur: "In a dramatic move, W3C CEO Jeff Jaffe has set fire to the W3C. Understandable; I've often wanted to do the same." |
| 22:01 | <Ms2ger> | zcorpan, http://www.whatwg.org/specs/web-apps/current-work/multipage/association-of-controls-and-forms.html#dom-fae-form is reflecting enough? |
| 22:02 | <zcorpan> | Ms2ger: that's not reflecting the content attribute |
| 22:05 | <annevk> | TabAtkins_: haha |
| 22:06 | <annevk> | mattur is on a role |
| 22:29 | <annevk> | "TheOtherAB" stuff keeps getting better |
| 22:51 | <Hixie> | so... interface A { void x(); void y(); }; is black-box identical to interface A { void x(); } [NoInterfaceObject] interface B { void y(); } A implements B; right? |
| 22:51 | <Hixie> | i think i ask this like every other day but for some reason i am never confident of the answer |
| 22:53 | <annevk> | you should use partial interface A { }; |
| 22:53 | <Hixie> | i am going to have multiple interfaces implement B |
| 22:53 | <Hixie> | so i can't do that |
| 22:54 | <annevk> | I think the answer is yes, but I'm also not sure :) |
| 22:54 | <annevk> | heycam: ^^ |
| 22:54 | <heycam> | Hixie, annevk, yep that's equivalent |
| 22:55 | <Hixie> | ok thanks |
| 22:57 | <Hixie> | heycam: can i have an 'implements' statement before the interface is declared? |
| 22:57 | <heycam> | Hixie, yes there's not really any ordering of definitions |
| 22:58 | <Hixie> | awesome |
| 23:02 | <Hixie> | ok, step one. extract the transformation methods and ensure everything still makes sense. |
| 23:02 | <Hixie> | done. |
| 23:04 | <annevk> | is someone on ietf-http-wg actually proposing to replace the single request model of http-aes with two requests? |
| 23:04 | <annevk> | I should probably stop reading that thread |
| 23:07 | <Hixie> | looks like i was dropped off the cc list |
| 23:08 | <annevk> | http://lists.w3.org/Archives/Public/ietf-http-wg/2012JanMar/0815.html is that one |
| 23:09 | <annevk> | http://lists.w3.org/Archives/Public/ietf-http-wg/2012JanMar/0816.html is one that seems to forget about not complicating the APIs |
| 23:11 | <Hixie> | oh i saw those e-mails |
| 23:11 | <Hixie> | didn't even realise they were talking about the same thing |
| 23:14 | <Hixie> | definitely didn't receive all the e-mails on that thread though |
| 23:16 | <Hixie> | in particular, looks like i didn't receive phk's e-mail... |
| 23:16 | <Hixie> | ah, it was marked spam |
| 23:17 | <Hixie> | looks like he doesn't understand the situation i described though |
| 23:17 | <Hixie> | oh well |
| 23:44 | <Hixie> | when we talk about converting text + a font specification to a specific set of drawing instructions, what's the term for that? |
| 23:44 | <Hixie> | rasterisation? |
| 23:45 | <Hixie> | i don't mean rendering, since it might not actually be shown, just turned into a path for further manipulation |
| 23:46 | <Philip`> | Rasterization would be converting to pixels |
| 23:46 | <gsnedders> | Vectorization? |
| 23:47 | <gsnedders> | But fonts are already vectors. |
| 23:47 | <gsnedders> | Well, except when they're bitmapped fonts. |
| 23:48 | <Hixie> | assume for the sake of this conversation that we can ignore bitmap fonts |
| 23:54 | <TabAtkins_> | Hm, I know the word, but it's not coming to mind right now. |
| 23:54 | <TabAtkins_> | Layout? |
| 23:55 | <heycam> | outlining |
| 23:55 | <heycam> | I think that's the term used in tools like Illustrator that take a text object and converts it to paths |
| 23:56 | <Hixie> | outlining might work, though that tends to imply "not filled in", which it could be |
| 23:57 | <Dashiva> | "Font rasterization is the process of converting text from a vector description (as found in scalable fonts such as TrueType fonts) to a raster or bitmap description." |
| 23:57 | <TabAtkins_> | Dashiva: Right, but that's not what Hixie wants. |
| 23:58 | <TabAtkins_> | He means doing the shaping, spacing, etc. necessary to get the final glyph vectors for the text, so you can then manipulate the vectors. |
| 23:58 | <Hixie> | (also, any ideas what Element should be used to determine the available fonts when you do some sort of font whateverthewordis on a Path object?) |
| 23:58 | <heycam> | the process of taking an input string, font selection, and coming up with a list of glyphs and their positions is probably just "layout" |
| 23:58 | <Hixie> | (should Path objects just be created with a Canvas owner for this purpose?) |
| 23:58 | <dbaron> | you shouldn't assume that that's the path implementations go through |
| 23:59 | <dbaron> | for example, they might rasterize each glyph of the font for the correct size and then blit the rasters |
| 23:59 | <Hixie> | heycam: dbaron we're talking about Path objects, where there's no rendering |
| 23:59 | <Hixie> | er |
| 23:59 | <Hixie> | dbaron ^ |
| 23:59 | <Hixie> | (my original question was "when we talk about converting text + a font specification to a specific set of drawing instructions, what's the term for that?") |