| 00:19 | <Benvie> | Domenic_: how stable is https://github.com/domenic/promises-unwrapping |
| 00:21 | <Benvie> | we're looking at implementing the rest of the API in JS while waiting for the C++ implementation (https://bugzilla.mozilla.org/show_bug.cgi?id=941920) |
| 00:27 | <Domenic_> | Benvie: it's done |
| 00:27 | <Benvie> | excellent |
| 00:27 | <Domenic_> | I might move some things into abstract operations for the module loader spec |
| 00:27 | <Domenic_> | Benvie: on that note see https://github.com/domenic/promises-unwrapping/blob/master/testable-implementation.js |
| 00:27 | <Benvie> | yeah I've looked through that. It's a good reference |
| 00:28 | <Domenic_> | without that this would have been madness |
| 00:28 | <Domenic_> | i'd make spec changes and get many test failures |
| 00:28 | <Benvie> | haha yeah |
| 00:28 | <Domenic_> | https://github.com/domenic/promises-unwrapping/issues/milestones |
| 00:28 | <Benvie> | I have a question about Promise.cast |
| 00:29 | <Benvie> | Promise.cast({ then: () => {}) |
| 00:29 | <Domenic_> | returns a forever-pending Promise |
| 00:29 | <Benvie> | it's not supposed to assimilate thenables, is this correct? |
| 00:29 | <Domenic_> | it is supposed to assimilate thenables |
| 00:30 | <Benvie> | oh yeah I see |
| 00:30 | <Benvie> | IsPromise will return false for it, but it will be assimilated as part of resolution |
| 00:30 | <Domenic_> | yeah |
| 00:30 | <Benvie> | thanks |
| 00:30 | <Domenic_> | The IsPromise is for memory-efficient casts when you know you have the right type |
| 00:31 | <Benvie> | yeah, it's just a shortcut |
| 00:31 | <Domenic_> | so if you are programming defensively you do deepFreeze(Promise); Promise.cast(thingy).then(...) |
| 00:31 | <Domenic_> | and you pay no memory costs |
| 00:31 | <Domenic_> | The constructor check is important because otherwise you could fool it with an ill-behaved subclass |
| 00:32 | <Benvie> | yeah. In our JS implementation we're not supporting subclasses so it cuts out some of the complexity there |
| 00:33 | <Domenic_> | yeah, probably all of it -_- |
| 00:33 | <Benvie> | haha yeah |
| 00:34 | <Domenic_> | spec was like done, without subclassing, on 2013-09-18 according to commit logs. Then I started getting envious of allenWB's Array tricksiness... |
| 00:35 | <Benvie> | it's good you did it though. If you hadn't, then it would have had to be added at some point anyway to get into ES6, since all the builtins have to be subclassable now |
| 00:37 | <Domenic_> | yeah, i like this new world of subclassable builtins |
| 00:37 | <Domenic_> | you guys should make array subclassable |
| 00:40 | <gsnedders> | Gah, just accidentally nuked my local logs of #whatwg going back years. Oh well. |
| 04:02 | <rniwa> | MikeSmith, jgraham: maybe either one of you could approve my PR to fix some html template tests? https://github.com/w3c/web-platform-tests/pull/442 |
| 04:02 | <rektide> | if i want to author some kind of spec, what should i do to get started? w3 seems to use this respec thing, but i'm not sure what it really consititutes, what it's for, what reading i need to do, how i should get started? |
| 04:03 | <rniwa> | rektide: my recommendation is to start writing the document somewhere |
| 04:03 | <rniwa> | rektide: and figure out those things later |
| 04:04 | <rniwa> | rektide: unless you've just signed up to be an editor of some specification |
| 04:06 | <rektide> | yeah... my urge to do it right is a hard one to ignore. i didn't really expect there being any onboarding guides or good ways to get started, but i figured i'd ask. |
| 04:07 | <rniwa> | rektide: have you read http://www.w3.org/html/wg/wiki/ExtensionHowTo ? |
| 04:08 | <rniwa> | rektide: I don't know what kind of specification you're intending to write |
| 04:08 | <rniwa> | rektide: but if it's something you haven't proposed somewhere yet |
| 04:08 | <rniwa> | rektide: then do that first |
| 04:08 | <rniwa> | rektide: it doesn't make much sense to put all your energy into making things right |
| 04:08 | <rniwa> | if your proposal has to change significantly due to feedback you get |
| 04:09 | <Domenic_> | rektide: I find GitHub + markdown specs is plenty. See e.g. https://github.com/whatwg/streams/ or even https://github.com/domenic/promises-unwrapping |
| 04:15 | <rektide> | rniwa: I am writing an extension atm. that ExtensionHowTo is an invaluable link, my sincere thanks |
| 04:15 | <rektide> | thanks both, thanks a dozen |
| 04:28 | rniwa | gets lunch |
| 09:04 | <hsivonen> | annevk-cloud: is there something wrong with my test case or do non-Gecko browsers not support responseType = "json" ? http://hsivonen.com/test/moz/json-bom/test.html |
| 09:28 | <odinho> | hsivonen: Presto does at least. |
| 09:28 | <odinho> | (if you tested that :P ) |
| 09:38 | <hsivonen> | odinho: I didn't |
| 09:38 | <odinho> | hsivonen: Okay. But now you know ;-) |
| 09:39 | <hsivonen> | odinho: boo. Presto supports UTF-16. |
| 09:40 | <odinho> | Yeah. Too bad there was no testharness.js test with that as a fail-condition when sof did that work. He's very responsive to fixing red tests. |
| 09:41 | <odinho> | Whenever I wanted him to fix something it was just to create a test for it, didn't have to say anything :P |
| 09:45 | <hsivonen> | I expected Blink to have responseType json support by now, but apparently not |
| 09:49 | <hsivonen> | hmm. I'm on 100 Mbps down connection and Presto Opera tells me I'm on a slow network |
| 09:52 | <odinho> | HEHE. Yes. |
| 09:52 | <zcorpan> | presto opera has high expectations |
| 09:52 | <odinho> | Opera 12 likes trolling. |
| 09:53 | <zcorpan> | 100 Mbps? Why not 100 Gbps? |
| 10:09 | <jgraham> | odinho: s/responsive fixing red tests/awesome/ |
| 10:10 | <jgraham> | hsivonen: Presto's slow network connection experience is one reason I am very skeptical of adding any kind of "low bandwidth detection" API |
| 10:18 | <smaug____> | hayato: hi. so about event propagation |
| 10:18 | <hayato> | smaug____: hi |
| 10:19 | <smaug____> | (I've been reviewing the patch for Gecko) |
| 10:19 | <hayato> | smaug____: sounds nice. |
| 10:20 | <smaug____> | I filed few bugs, but those are quite clear, I think |
| 10:20 | <smaug____> | but 6.2, Event paths |
| 10:20 | <hayato> | smaug____: what's problem? |
| 10:20 | <smaug____> | there is some strangeness |
| 10:20 | <smaug____> | putting all the insertion points to the path |
| 10:21 | <smaug____> | that leads to a setup where the path may contain some elements, but not their parent nodes |
| 10:21 | <smaug____> | even though element and its parent belong to the same shadow tree |
| 10:21 | <smaug____> | what is the reason to put all the insertion points to the path |
| 10:21 | <smaug____> | and not only the final destination |
| 10:24 | <hayato> | smaug____: I don't have a strong opinion for that. I guess some developers want to register listeners to an insertion points, which is not the final destination to monitor a distributed node. |
| 10:24 | <hayato> | But I am feeling you are right. |
| 10:25 | <hayato> | smaug____: The final destination is enough for most use cases. |
| 10:25 | <smaug____> | yup |
| 10:25 | <hayato> | AFAIR, no one has expressed opinion for this part of the spec. |
| 10:26 | <hayato> | smaug____: I think we can change the part of the spec. |
| 10:26 | <smaug____> | it is mainly the consistency of event target chain which breaks if also non-final destination is put to the path |
| 10:26 | <smaug____> | hayato: good! |
| 10:27 | <hayato> | smaug____: Okay. Let me think again. |
| 10:27 | smaug____ | likes consistency and simplifying things |
| 10:28 | <hayato> | smaug____: Let me do that in a few business days. |
| 10:28 | <smaug____> | k |
| 10:29 | <smaug____> | thanks |
| 10:30 | <smaug____> | hayato: I could file a bug |
| 10:30 | <odinho> | jgraham: sof is moved to blink now, if you didn't know :) |
| 10:30 | <hayato> | smaug____: for the spec? I appreciate that! |
| 10:31 | <smaug____> | yeah |
| 10:31 | <hayato> | smaug____: thanks! |
| 10:31 | <smaug____> | hayato: what is your w3c bugmail? |
| 10:31 | <jgraham> | odinho: I did. I have high hopes for blink suddenly becoming much more spec compliant :) |
| 10:31 | <smaug____> | looks like components stuff defaults to dglazkov |
| 10:31 | <hayato> | smaug____: hayato⊙co |
| 10:32 | <odinho> | jgraham: Haha. Lots of pressure on the man! |
| 10:32 | <hayato> | smaug____: I don't know how to change the default. :) |
| 10:32 | <jgraham> | odinho: No pressure, but seriously, he seems to eat bugs for breakfast ;) |
| 10:33 | <odinho> | this is true |
| 10:34 | <darobin> | the great thing with eating bugs for breakfast is that they don't get all soggy in the milk |
| 10:34 | <smaug____> | hayato: does dglazkov handle still some of the web components stuff? |
| 10:35 | <smaug____> | or should all the bugs be assigned to you by default ? |
| 10:35 | <smaug____> | I'm sure MikeSmith could ask someone to change the default, if needed |
| 10:37 | <hayato> | smaug____: Ops. I am afraid that we don't have a category for Shadow DOM. There is only for ComponentModel which includes Shadow DOM. |
| 10:37 | <Ms2ger> | Note to self: don't typo firewall as fireball |
| 10:38 | <hayato> | smaug____: dglazkov absolutely handles a lot of web components stuff. So the current default is okay. |
| 12:23 | <hsivonen> | these days Web authors are so clueful that it's hard to find an unlabeled non-windows-1252 site for testing |
| 12:38 | <hsivonen> | where do I find an unlabeled non-windows-1252 site on a non-windows-1252-affiliated country TLD? |
| 12:38 | <hsivonen> | there are tons of these, but now that I want to find one, where are they? |
| 12:38 | <jgraham> | hehe |
| 13:45 | <zcorpan> | hsivonen: what are the non-windows-1252-affiliated TLDs? |
| 13:49 | <zcorpan> | jgraham: was there any conclusion about how to do cleanup for async tests? |
| 13:50 | <hsivonen> | zcorpan: https://bugzilla.mozilla.org/attachment.cgi?id=8336745&action=diff#a/dom/encoding/domainsfallbacks.properties_sec2 |
| 13:50 | <zcorpan> | jgraham: i think it might be good with a 'result' callback on the Test itself |
| 13:54 | <jgraham> | zcorpan: Yeah |
| 13:54 | <jgraham> | Could certainly have a "finally" function |
| 13:55 | <jgraham> | Not sure what the syntax should be, exactly though |
| 13:55 | <zcorpan> | jgraham: i think it should be possible to define the function inside a step. possibly be able to define several functions that will be invoked (like event listeners) |
| 13:56 | <zcorpan> | jgraham: in order to use variables in that scope |
| 13:56 | <jgraham> | zcorpan: I guess |
| 13:56 | <jgraham> | Seems complicated |
| 13:56 | <jgraham> | But it could work I guess as this.add_cleanup(function(){}) |
| 13:56 | <zcorpan> | yeah |
| 13:58 | <zcorpan> | so if i create an iframe and append it to the document, i can define the cleanup function right next to it and not have to store the iframe in a global variable or anything |
| 13:59 | <jgraham> | zcorpan: I'm on it |
| 13:59 | <zcorpan> | cool |
| 14:08 | <jgraham> | zcorpan: I guess the cleanup ought to be called even if the test doesn't finish running? |
| 14:08 | <jgraham> | I mean, if it gets status NOT RUN |
| 14:09 | <darobin> | yes |
| 14:09 | <zcorpan> | is that the same as TIMEOUT? or isn't TIMEOUT a thing anymore? |
| 14:09 | <darobin> | I reckon cleanup should always run no matter what |
| 14:09 | <jgraham> | (at that point the user will navigate the page ofc, but it's possible that the test will do something that will persist across navigation) |
| 14:09 | <jgraham> | TIMEOUT is a thing still |
| 14:10 | <zcorpan> | my thinking was that cleanup would be invoked at the same time as 'result' |
| 14:10 | <jgraham> | But maybe NOT RUN only happens if you don't run any steps these days; I gorget how the semantics ended up |
| 14:11 | <MikeSmith> | hayato: if you want a Shadow DOM component created, we can can make one |
| 14:21 | <hsivonen> | MikeSmith: FYI, validator.nu and html5.validator.nu now run on OpenJDK 7 |
| 14:22 | <MikeSmith> | hsivonen: ah cool |
| 14:25 | <hsivonen> | migrated to Ubuntu 12.04 just in time for 14.04 |
| 14:26 | <MikeSmith> | so I hope Java versioning is stable at 7 now for a while |
| 14:26 | <MikeSmith> | I've not found anything that was added in 7 that I have any compelling need for |
| 14:27 | <MikeSmith> | so I can't imagine that anything added in the next major version will be anything really useful |
| 14:30 | <hsivonen> | MikeSmith: Java 7 added SNI |
| 14:30 | <hsivonen> | MikeSmith: though I haven't tested if the old HttpClient uses it |
| 14:30 | <MikeSmith> | I have to admit I don't know what SNGI is |
| 14:30 | <MikeSmith> | *SNI |
| 14:30 | <hsivonen> | MikeSmith: that's gonna be necessary when IE for XP dies and https virtual hosting takes off |
| 14:30 | <MikeSmith> | oh |
| 14:30 | <hsivonen> | MikeSmith: it's like the HTTP Host header but for TLS |
| 14:31 | <MikeSmith> | Server Name Indication |
| 14:31 | <MikeSmith> | I've never heard of it til now |
| 14:31 | <hsivonen> | MikeSmith: because the IE/XP boat anchor means it can't be deployed |
| 14:32 | <hsivonen> | though first, the Android 2.x boat anchor needs to fade away, too |
| 14:33 | <hsivonen> | it's shameful how Google shipped Anroid 2.x without SNI |
| 14:35 | <hsivonen> | MikeSmith: sadly, validator.nu even on OpenJDK 7 doesn't support SNI |
| 14:35 | <hsivonen> | gotta fix that! |
| 14:36 | <jgraham> | zcorpan: https://critic.hoppipolla.co.uk/r/436 or something? |
| 14:37 | <jgraham> | I wonder what happens if window.onerror is called |
| 14:38 | <jgraham> | I guess it still works |
| 14:39 | <zcorpan> | hsivonen: i have a list of 276 urls with those tlds that don't have charset in content-type or <meta> |
| 14:40 | <hsivonen> | zcorpan: cool. can you post the list somewhere? |
| 14:40 | <zcorpan> | yep |
| 14:45 | <zcorpan> | hsivonen: i didn't exclude ascii-only pages, but you could filter for that if you download the data set |
| 14:46 | <hsivonen> | zcorpan: ok |
| 14:46 | hsivonen | learns that there's a new Java dependency management tool https://ant.apache.org/projects/ivy.html |
| 14:47 | <hsivonen> | "People don't like Maven. Let's name the replacement to something that people associate with poison. That'll drive adoption." |
| 14:51 | <zcorpan> | jgraham: yep. thanks! |
| 14:54 | <jgraham> | zcorpan: My only concern about that design is that it doesn't have great atomicity properties |
| 14:55 | <jgraham> | Although it's hard to think of a design that does |
| 14:55 | <jgraham> | e.g. if you do |
| 14:55 | <jgraham> | set_some_state() |
| 14:55 | <jgraham> | set_some_more_state() |
| 14:55 | <jgraham> | this.add_cleanup(rollback_state) |
| 14:56 | <jgraham> | and set_some_more_state fails, you won't clean up the first lot of state |
| 14:56 | <jgraham> | (those are only functions for convenience, they could be statements or whatever) |
| 14:56 | <zcorpan> | yeah, then you'd have to use two add_cleanups |
| 14:56 | <jgraham> | Right, but people clearly aren't going to do that. I guess it is just something we will have to live with |
| 14:56 | <zcorpan> | which i noticed i didn't do in my test that inserts two elements to the document |
| 14:57 | <zcorpan> | though they both append to body so if the first one succeeds the second one really should also |
| 14:57 | <zcorpan> | yeah |
| 14:58 | <jgraham> | OK, let's see how that works out |
| 14:58 | <jgraham> | Maybe we can change the API if it is only used a little bit and turns out to be too hard to use |
| 14:59 | <zcorpan> | my first impression is that it's awesome :-) |
| 14:59 | <jgraham> | Well that is good at least :) |
| 14:59 | <darobin> | the only question I had was about whether there should be a specific context for the callback |
| 14:59 | <darobin> | but I guess that's nothing that bind() can't do for you |
| 14:59 | <zcorpan> | even for sync test it looks better than try/finally because the cleanup can be defined next to the relevant thing |
| 15:00 | <jgraham> | darobin: I don't think there is anything that makes sense by default? |
| 15:00 | <darobin> | jgraham: that was my second thought |
| 15:00 | <jgraham> | But it could take a this_obj ofc |
| 15:01 | <hsivonen> | occasionally I wonder if the HttpClient devs are just trolling downsteam users by refactoring the API in every release |
| 15:01 | <darobin> | yeah, I thought about that — but it's adding wiring for stuff that people can do for themselves easily |
| 15:01 | <darobin> | other than that, it means I should probably update my docs |
| 15:02 | <zcorpan> | let's keep it simple for now |
| 15:04 | <darobin> | +1 |
| 15:11 | <zcorpan> | jgraham: what should i do to get wptserve running again? :-) |
| 15:12 | <jgraham> | zcorpan: Gah |
| 15:12 | <jgraham> | I wonder if I can get my hands on a Mac |
| 15:13 | <Ms2ger> | Look around the office; about half of the people in your line of sight will have some kind of macbook :) |
| 15:15 | <zcorpan> | i checked out the previous commit for now |
| 15:27 | <hsivonen> | I'm sure this new HttpClient API makes sense to its creators... |
| 15:29 | <hsivonen> | old versions had *documentation* |
| 15:36 | <zcorpan> | has anyone researched how many attributes elements have in the wild? |
| 15:38 | zcorpan | finds http://canvex.lazyilluminati.com/misc/stats/attrcounts.png |
| 15:40 | <Ms2ger> | zcorpan, svg is rumoured to have lots |
| 15:42 | <zcorpan> | Ms2ger: but svg isn't very common |
| 15:43 | <Ms2ger> | True |
| 15:46 | <zcorpan> | i'm writing some tests for resolving urls, do you guys prefer if i open a PR now when it's half-finished or should i wait until i think it's complete? |
| 15:47 | <Ms2ger> | Feel free to do it now |
| 15:48 | <zcorpan> | k |
| 15:56 | <zewt> | gar @ pages that redirect to /maintenance when offline |
| 16:05 | <zcorpan> | https://critic.hoppipolla.co.uk/r/437 |
| 16:07 | <zcorpan> | jgraham: this reminded me that testharness.js is annoying for utf-16 tests, i need to use <link charset> |
| 16:08 | zcorpan | *poof* |
| 16:08 | <jgraham> | Sigh |
| 16:09 | <jgraham> | zcorpan is now the second person in a row who had disappeared as I was typing a message for them |
| 16:10 | <Ms2ger> | Probably should make it utf-8 in http |
| 16:12 | <jgraham> | Yeah, but I don't quite know the right way to do that |
| 16:12 | <jgraham> | Some sort of configuration file? |
| 16:12 | <jgraham> | Custom magic in the server? |
| 16:12 | <jgraham> | I guess custom magic isn't so bad |
| 17:42 | <jgraham> | Achievement unlocked: person uses sync XHR as a justification foradding more blocking APIs |
| 18:09 | jensnockert | hugs jgraham. |
| 18:36 | <wilhelm_> | zcorpan, the most popular guy at the party: http://testthewebforward.org/assets/blog/shenzhen//testtwf-shenzhen25.jpg |
| 18:37 | <Ms2ger> | Is he the leftmost guy? |
| 18:43 | <miketaylr> | heh |
| 18:43 | <miketaylr> | (that's leif arne) |
| 18:52 | <Hixie> | anyone got any idea what kind of diagram would help with MessagePort objects, Window.postMessage, and so forth? https://www.w3.org/Bugs/Public/show_bug.cgi?id=23851 |
| 18:59 | <Hixie> | can anyone find the e-mail address of 'Robert D. Thompson, author of "The MFC Programmer's Reference" (1997)' ? |
| 19:03 | <miketaylr> | Hixie: found his linkedin, http://www.linkedin.com/in/robertdeanthompson |
| 19:03 | <Hixie> | any way to communicate with him from there? |
| 19:03 | <Hixie> | (i don't do linked in) |
| 19:04 | <miketaylr> | i guess you can send "inmail", i've never used it |
| 19:04 | <miketaylr> | also https://plus.google.com/105991459354415331406/posts |
| 19:04 | <miketaylr> | but i dont do g+ :p |
| 19:05 | <Hixie> | i can use g+, thanks |
| 19:05 | <Hixie> | (though it looks like he doens't actively use it) |
| 19:07 | <miketaylr> | meh, have to upgrade to even send InMail™ |
| 19:08 | <Hixie> | g+ isn't letting me +him into a post |
| 19:08 | <Hixie> | ok |
| 19:08 | <Hixie> | i guess i can send him a hangouts message but he's not online and probably won't get it for months if he doesn't use hangouts |
| 19:08 | <Hixie> | i could send him a youtube message but he won't get that since he doesn't seem to use youtube |
| 19:08 | <Hixie> | oh well |
| 19:14 | <marcosc_> | that was the moment zcorpan figured out we could do <picture> another way |
| 19:16 | <TabAtkins> | Are other people actually capable of seeing that image? It sounds like you are, but I get a 404. |
| 19:44 | <SimonSapin> | TabAtkins: you mean testtwf-shenzhen25.jpg? I see it |
| 20:15 | <Ms2ger> | OH: "Use a bug in the UTF-7 decoder..." |
| 20:20 | <TabAtkins> | SimonSapin: Hm. I click on that link, which navigates to http://testthewebforward.org/assets/blog/shenzhen//testtwf-shenzhen25.jpg, which 404s. |
| 20:26 | <Hixie> | what browser? |
| 20:26 | <miketaylr> | 200 OK |
| 20:26 | <Hixie> | http://testthewebforward.org/assets/blog/shenzhen//testtwf-shenzhen25.jpg works for me in firefox and chrome and... i'd test webkit but my webkit seems to have died. |
| 20:26 | <Hixie> | brb fixing webkit. |
| 20:28 | <Hixie> | wfm in webkit too |
| 20:29 | <Hixie> | you sure you're not getting a trailing comma or something? |
| 20:35 | <TabAtkins> | Yes, quite sure there's no comma. |
| 20:35 | <TabAtkins> | Note that the comma isn't underlined here, and there's no comma on the previous one at all. |
| 20:35 | <TabAtkins> | And this is on my Pixel, recently updated. |
| 20:36 | <TabAtkins> | Ms2ger: The presence of a UTF-7 decoder is the bug, right? |
| 20:41 | <Hixie> | TabAtkins: weird! does it work if you change // to / ? |
| 20:41 | <TabAtkins> | Nope! |
| 20:42 | <TabAtkins> | That was the first thing I thought was wrong - I assumed a typo. |
| 20:42 | <Hixie> | woah, now i'm getting a 404 |
| 20:42 | <Hixie> | wtf |
| 20:42 | <Hixie> | i definitely didn't have it cached before |
| 20:42 | <Hixie> | ohhhhh hey |
| 20:43 | <Hixie> | http://testthewebforward.org/ gives me a temporary error |
| 20:43 | <SimonSapin> | http://i.imgur.com/V8JgIAN.jpg |
| 20:43 | <Hixie> | i bet there's round-robin DNS going on here |
| 20:44 | <TabAtkins> | Thanks, SimonSapin! |
| 20:45 | <Hixie> | i get 204.232.175.78 as the IP when it works |
| 20:45 | <Hixie> | 2001:470:0:cb::4113:bb64 when it doesn't |
| 20:46 | <Hixie> | same DNS server, though |
| 20:46 | <Hixie> | wonder why it gives different results sometimes |
| 20:46 | <Hixie> | anyway, looks like testthewebforward.org has badly configured IPv6 |
| 20:46 | <TabAtkins> | Weird, ok. |
| 20:46 | <Hixie> | and the DNS server sometimes returns an AAAA record |
| 22:02 | <TabAtkins> | Oh man, Chrome is telling me which tab is making noise now. |
| 22:48 | <Hixie> | TabAtkins: you around? is there some way to say that an element should act like an inline-block even when it's set to other display types, for the purposes of sizing, while still allowing the other display types to remain relevant for other reasons (e.g. participating in the table model, if set to display:table-cell)? |
| 22:49 | <Hixie> | hm, actually, that's not what's going on here... |
| 22:52 | <Hixie> | http://software.hixie.ch/utilities/js/live-dom-viewer/saved/2650 |
