| 00:30 | GPHemsley | wonders what to do with http://wiki.whatwg.org/wiki/AAFC_Metadata_Application_Profile |
| 06:41 | <Ms2ger> | zewt, test succeeded :) |
| 06:42 | <Ms2ger> | The issue is that checking for exceptions at every step is terribly error-prone |
| 06:54 | <zcorpan> | this dekiss guy is adding a lot of noise |
| 07:08 | <zcorpan> | can we make anolis warn about xref-able elements without a title that don't get xreffed? |
| 07:12 | <Ms2ger> | Actually, you can make it error |
| 07:13 | <Ms2ger> | --use-strict |
| 08:47 | <codeho> | argh |
| 08:58 | <smaug____> | annevk: ping |
| 08:58 | <annevk> | smaug____: whoa, why are you awake? |
| 08:58 | <smaug____> | (someone called me around 9am and tried to sell something) |
| 08:59 | <smaug____> | and hey, it is noon |
| 08:59 | <Ms2ger> | smaug____, exactly :) |
| 08:59 | <smaug____> | annevk: https://bugzilla.mozilla.org/show_bug.cgi?id=780953 |
| 08:59 | <smaug____> | do you recall if that has been discussed somewhere |
| 08:59 | <smaug____> | whatwg or w3 |
| 09:00 | <annevk> | smaug____: hasn't been discussed |
| 09:00 | <annevk> | smaug____: pretty sure anyway, I recommend raising on WHATWG |
| 09:01 | <annevk> | I don't think OS-level changes have had much consideration in general |
| 09:02 | <darobin> | smaug____: this has been discussed in sysapps |
| 09:03 | <smaug____> | darobin: well, needs to happen in WhatWG, at least partially |
| 09:03 | <darobin> | smaug____: https://www.w3.org/Bugs/Public/show_bug.cgi?id=21290 |
| 09:03 | <smaug____> | since the change requires Navigator to inherit EventTarget |
| 09:03 | <smaug____> | ah |
| 09:03 | <smaug____> | thanks |
| 09:03 | <darobin> | it needs to happen in the HTML spec for sure I would say |
| 09:04 | <smaug____> | uh, could I change the product |
| 09:04 | <darobin> | who does it, I don't care |
| 09:04 | <darobin> | if you want to put it on the WHATWG's plate, please don't change the product, but instead clone it |
| 09:04 | <darobin> | I reckon it'll happen anyway, it's useful |
| 09:06 | <smaug____> | yeah, it is useful |
| 09:06 | <smaug____> | and simple thing |
| 09:06 | <darobin> | indeed |
| 09:07 | <zcorpan> | MikeSmith: https://dvcs.w3.org/hg/csswg/rev/20c14a9da2c1 gives me 500 |
| 09:10 | <zcorpan> | Ms2ger: thanks. however, it gives an error for <!--copyright--> :-( |
| 09:10 | <Ms2ger> | Lovely |
| 09:11 | Ms2ger | puts that on his todo list |
| 09:13 | <zcorpan> | probably <abbr> shouldn't be an xref element |
| 09:14 | <annevk> | smaug____: it's already filed against WHATWG, darobin referenced the wrong bug |
| 09:14 | <annevk> | smaug____: https://www.w3.org/Bugs/Public/show_bug.cgi?id=21289 |
| 09:14 | <Ms2ger> | Yeah, I think it's based on Bert's xref elements |
| 09:15 | <zcorpan> | i'd like it if <var> wasn't an xref element, too |
| 09:15 | <annevk> | Ms2ger: should move anolis to GitHub |
| 09:15 | <annevk> | yeah, we should not have <var> be xref |
| 09:15 | <annevk> | or <i> |
| 09:15 | <zcorpan> | why not <i>? |
| 09:15 | <Ms2ger> | annevk, you know I'm not much of a git fan :) |
| 09:16 | <annevk> | zcorpan: <i> is for algorithm flags and they're typically not xreffed |
| 09:16 | <annevk> | Ms2ger: that's a losing proposition |
| 09:17 | <Ms2ger> | Some people like to think so |
| 09:18 | darobin | didn't reference the wrong bug, just the bug he knew of (which is also the correct one ;) |
| 09:18 | <Ms2ger> | darobin, not in this channel ;) |
| 09:19 | <darobin> | zcorpan: are you still getting 500 on dvcs.w3? it seems to be back up here |
| 09:19 | <zcorpan> | darobin: ah, works now |
| 09:20 | darobin | likes <a> being an xref element, strangely |
| 09:22 | <smaug____> | annevk: ah, thanks |
| 09:22 | <annevk> | darobin: instead of <span>? |
| 09:22 | <annevk> | darobin: that might actually be nice |
| 09:23 | <annevk> | Ms2ger: ^^ |
| 09:23 | <Ms2ger> | File a bug :) |
| 09:23 | <darobin> | annevk: that's what ReSpec uses |
| 09:24 | <darobin> | it makes more sense I find |
| 09:24 | <darobin> | it's even a correct use of HTML!!!!1 |
| 09:37 | <zcorpan> | Ms2ger: https://bitbucket.org/ms2ger/anolis/pull-request/6/remove-and-from-the-list-of-xref-elements/diff |
| 09:38 | <jgraham> | darobin: Don't be silly, it's not possible to use HTML correctly. |
| 09:41 | <darobin> | jgraham: I was waiting for that one :) |
| 09:42 | <jgraham> | We have a great future as a comedy double act. |
| 09:44 | <Ms2ger> | Dry British wit... And darobin? |
| 09:48 | <darobin> | Drunk French humour? |
| 09:48 | <Ms2ger> | Sounds about right |
| 09:48 | Ms2ger | passes darobin a glass of wine |
| 09:51 | <zcorpan> | should i add <a> in the same PR? |
| 09:51 | darobin | puts on a béret and camembert sandwich |
| 09:52 | <zcorpan> | i guess <a> would need more work than just adding it to the list to not screw up |
| 09:54 | <Ms2ger> | Sure |
| 09:54 | <Ms2ger> | (Did you run tests?) |
| 09:55 | <zcorpan> | no, i didn't |
| 09:55 | <zcorpan> | FAILED (failures=1) |
| 09:57 | <zcorpan> | FAIL: test_tests/toc-basic.src.html, however that seems unaffected by the change |
| 10:06 | <MikeSmith> | zcorpan: I been getting reports of intermittent 500s from dvcs.w3.org but can't see any problem on the server side |
| 10:06 | <zcorpan> | MikeSmith: ok |
| 10:07 | <MikeSmith> | the only troubleshooting solution that's worked in the past is for use to just restart apache there |
| 10:07 | <MikeSmith> | but in the past the only time we had 500s was when the server was completely wedged and consistently unresponsive |
| 10:36 | <joesavage> | Unsure if this is the right place to ask, but an element on my page (an "a" element with an "img" inside) simply isn't being focused in the Tab cycle, how can I fix this? I've tried adding a "tabindex" value, but it doesn't work, instead the textbox just next to it in the markup always gets focused first, and then everything else on the page after that. |
| 10:37 | <darobin> | joesavage: perhaps not the best forum for such questions, but does your <a> have an href with something useful in it? |
| 10:37 | <darobin> | if not, it's not focusable |
| 10:39 | <joesavage> | It's just linking to "/", but I've tried a bunch of other values in my troubleshooting process, so I guess it's not this. |
| 10:40 | <joesavage> | Interestingly, if I add some text after the image (i.e. <a href="/"><img src="whatever"/>test</a>), then it focuses first. |
| 10:42 | <joesavage> | I also notice that taking the "float: left;" away from it in the CSS makes it focus correctly too. Very odd. |
| 11:58 | <annevk> | fwiw: http://annevankesteren.nl/2013/06/london-tag (not comprehensive) |
| 12:03 | <jgraham> | annevk: "Getting there from where we are today is though" |
| 12:03 | <Ms2ger> | Sounds right |
| 12:04 | <jgraham> | Also, I'm not sure why the "champions" model is a good thing |
| 12:05 | <annevk> | jgraham: not sure what you meant by quoting that? |
| 12:05 | <Ms2ger> | tough? |
| 12:06 | <jgraham> | You either misspelt or just missed a word |
| 12:06 | <annevk> | ooh |
| 12:06 | <Ms2ger> | Damn English |
| 12:07 | <annevk> | jgraham: thanks |
| 12:07 | <annevk> | jgraham: not sure if champions is good, but is there anything else that scales? |
| 12:11 | <tobie> | annevk: why a separate GitHub org for W3C tag? |
| 12:12 | <Ms2ger> | See discussion in this channel a few days ago |
| 12:14 | <annevk> | tobie: easier to setup repos |
| 12:14 | <tobie> | annevk: how so? |
| 12:15 | <tobie> | I mean, if TAG members can't get admin rights on gh.com/w3c... |
| 12:15 | <tobie> | we'd need to fix that. :) |
| 12:16 | <tobie> | annevk: feels like a missed opportunity for visibility and transparency. |
| 12:17 | darobin | thinks it doesn't make that much difference |
| 12:17 | <annevk> | that sounds very opaque |
| 12:34 | <tobie> | annevk: what does? |
| 13:39 | <zewt> | Ms2ger: not explicitly saying what to do with exceptions is terribly not saying what to do--if you don't know exactly which steps in an algorithm might throw an exception, you can't see all of the code paths |
| 13:40 | <zewt> | and if the spec author doesn't know all of those places, that's error prone as well |
| 14:11 | <annevk> | tobie: what you said about the TAG not sharing a GitHub account |
| 14:12 | <tobie> | what's opaque about that? |
| 14:14 | <annevk> | it seems important our work is on GitHub, it doesn't matter much where afaict |
| 14:15 | <annevk> | and in fact, for people interested in what the TAG works on, a single page seems a lot clearer |
| 14:15 | <jgraham> | No one cares about that though |
| 14:16 | <jgraham> | Or no one should |
| 14:16 | <jgraham> | They should care about "are good web technologies being standardised and shipped" |
| 14:18 | <jgraham> | (a similar argument could be made that the CSS WG needs it own repos, and the HTML WG and the WebApps WG, and the Web Performance working group and…) |
| 14:18 | <jgraham> | s/repos/account/ |
| 14:20 | <annevk> | I think it helps to have some granularity so you know who is accountable |
| 14:20 | <annevk> | That we have a gazillion groups doing APIs these days is indeed not helping |
| 14:21 | <gsnedders> | But organizations don't do work; individuals do. |
| 14:21 | <jgraham> | I don't understand your argument |
| 14:22 | <jgraham> | Even if we only had HTML/WebApps/CSS it would still be strictly worse to split up all their stuff |
| 14:22 | <jgraham> | If you want namespacing that is possible by naming the repos groupname-whatever |
| 14:22 | gsnedders | doesn't really know what's going on, being on holiday and all |
| 14:25 | <jgraham> | (c.f. Conway's Law) |
| 14:39 | <annevk> | jgraham: I see it more as being about size |
| 14:39 | <annevk> | jgraham: if it's too large it becomes hard to follow |
| 14:40 | <annevk> | jgraham: e.g. that's why some specs have their own twitter account, or why we have separate specs to begin with |
| 14:40 | <annevk> | jgraham: even though they're all representing the same platform |
| 14:40 | <darobin> | I thought Conway's law was if you arrange black squares on a grid in a special way, you get spaceships |
| 14:41 | <annevk> | want |
| 14:42 | <SimonSapin> | sounds accurate |
| 14:42 | <Ms2ger> | Fun fact: the game of life is Turing-complete |
| 14:42 | Ms2ger | found that part of that textbook more interesting than the part about RDF |
| 14:43 | <jgraham> | Fun fact: So is W3C Process |
| 14:44 | <jgraham> | Dammit I think I just nerd-sniped myself. Now I wonder if you can abuse W3C Process to perform computations |
| 14:44 | <Philip`> | Implementing a Turing machine in Life is unfortunately slightly less practical than implementing one in C++ templates |
| 14:44 | <darobin> | jgraham: ROFL |
| 14:45 | <darobin> | well, we have loops for sure |
| 14:45 | <darobin> | I don't think you need the full process, Rec track might be T-complete |
| 14:45 | darobin | wonders what textbook Ms2ger had that featured both the GoL and RDF |
| 14:46 | <darobin> | someone must've been smoking something pretty damn good at the textbook factory |
| 14:46 | <Ms2ger> | darobin, Rich's Automata, Computability and Complexity |
| 14:47 | <Ms2ger> | It also featured claims that HTML is a regular language |
| 14:47 | <darobin> | I guess it was more than just smoking then |
| 14:48 | <darobin> | mmm, that book ain't cheap, too |
| 14:48 | <Ms2ger> | Indeed it isn't |
| 14:51 | <darobin> | I reckon that if you treat drafts as storage space then given LC is a conditional loop you have Turing-completeness |
| 14:53 | <gsnedders> | Unforunately it doesn't define the computation engine itself, merely everything else. |
| 14:54 | <darobin> | :) |
| 16:33 | <jgraham> | So, just how many testing-related IRC channels does W3C need? |
| 16:34 | <Ms2ger> | 42 |
| 16:35 | <jgraham> | That would explain a lot |
| 16:36 | <tobie> | jgraham: #htmlt, #test, #testing what else? |
| 16:47 | <jgraham> | tobie: I have no idea. Until today I didn't know that #test existed. |
| 16:47 | <MikeSmith> | me neither |
| 16:47 | <MikeSmith> | what's #test for? |
| 16:47 | <tobie> | jgraham: with the move to GitHub, it feels like some reconciling would be useful. |
| 16:49 | <tobie> | jgraham: including mailing lists |
| 16:49 | <tobie> | tracking that here: |
| 16:49 | <tobie> | https://github.com/w3c/ttwf-docs/issues/7 |
| 16:49 | <tobie> | MikeSmith: used by the testing tf |
| 16:50 | <MikeSmith> | ok |
| 16:50 | <Ms2ger> | #css-test |
| 16:56 | <tobie> | Ms2ger: ty |
| 16:56 | <Ms2ger> | Np |
| 18:17 | <jgraham> | tobie: Anyway I would happily close them all apart from #testing |
| 18:17 | <jgraham> | Mind you, I wouldn't spin up a bunch of closed membership task forces either |
| 18:23 | <tobie> | jgraham: neither would I, but corps are touchy when it comes to $$$. |
| 18:32 | <dekiss> | how is html language made? who is making it? html editors? |
| 18:32 | <dekiss> | I have hard time understanding it, how that mailing lists work etc. |
| 18:37 | <jgraham> | Unrelatedly, is there some reason that people are pushing everything to happen at "end of microtask" rather than just using normal tasks? |
| 18:38 | <jgraham> | (for small values of "everything" that includes a bunch of new stuff) |
| 18:53 | <Ms2ger> | So we're at the point where we're doing d3e conf calls without anything to talk about again? |
| 19:00 | <jgraham> | Someone else is doing D3E calls |
| 19:12 | <jgraham> | Also, "we were forced to do bad thing $X because of the $$$" seems to be the standard W3C excuse at the moment |
| 19:18 | <jarek> | is there any chance that Goolge will be trying to standardise DOM improvements made in Dart? |
| 19:19 | <jarek> | I mean things like query() or events as streams |
| 19:19 | <Ms2ger> | The idea behind those was that there's no legacy dart content, so breaking stuff is free, right? |
| 19:19 | <Ms2ger> | Breaking stuff on the actual web is far from free |
| 19:21 | <jarek> | I'm not aware of Dart internals, but it looks like most of the improvements are implemented as wrappers on top of standard DOM |
| 19:21 | <jarek> | http://www.dartlang.org/articles/improving-the-dom/ |
| 19:22 | <tantek> | jgraham - w3cexcuses.tumblr.com ? |
| 19:23 | <jgraham> | jarek: Nor am I, but I doubt they want to reimplement the C++ bits |
| 19:24 | <jgraham> | I mean, the DOM still has to exist. So having something fundamentally different exposed to Dart and JS seems hard |
| 19:25 | <jarek> | why NodeList and DOMTokenList don't inherit from Array.prototype? Does the spec prohibit this explicitly? |
| 19:25 | <Ms2ger> | NodeList explicitly inherits from Array.prototype in the spec |
| 19:25 | <Ms2ger> | We're waiting on Blink to try implementing it |
| 19:25 | <Ms2ger> | Last time we tried, a lot of Google sites broke |
| 19:26 | <Ms2ger> | And having wrappers on top of the DOM only increases the API surface for little gain, making the platform harder to learn and to maintain |
| 19:27 | <jgraham> | Ms2ger: Wait, what? Why does the spec say something that's known not to work? |
| 19:28 | <annevk> | TabAtkins: yo, if you're around, if you could reply to the www-dom thread with MM that'd be grand |
| 19:28 | <Ms2ger> | jgraham, because Google could maybe have fixed its site |
| 19:28 | <Ms2ger> | s |
| 19:28 | <jgraham> | It should at best say Note: It has been suggested that (foo). Experiments with (foo) showed a significant web compat impact. |
| 19:28 | <jgraham> | Ms2ger: If it broke Google sites it seems highly likely it also broke other sites |
| 19:29 | <Ms2ger> | Perhaps |
| 19:29 | <Ms2ger> | I haven't looked into it closely |
| 19:29 | <jgraham> | and expecting Google to spend resources fixing its sites to make some blink people happy seems unreasonably optimistic |
| 19:30 | <jgraham> | In any case I don't think the spec should be left in a state where it requires knwon-broken behaviour |
| 19:30 | <jgraham> | *known |
| 19:41 | <Ms2ger> | jgraham, notice added |
| 19:43 | <jgraham> | Ms2ger: Thanks |
| 20:37 | <GPHemsley> | Hixie: Ah, one catch you forgot to mention about keeping the whole spec open: I have to reload it all when changes are made! ;) |
| 21:52 | <dglazkov> | good morning, Whatwg! |
| 23:18 | <dekiss> | can I send informations from server to client every 1 min while web page is open? websockets I think is used for this? is that ok if I use it or which are my options? |
| 23:25 | <Hixie> | dekiss: for web authoring questions you probably want another forum, like #html5 or some such |
| 23:25 | <Hixie> | dekiss: here you're more likely to get snarky comments and venting than advice :-) |
| 23:25 | <dekiss> | can I restrict some web page so not all can see it? and if yes how please? |
| 23:26 | <dekiss> | should I check for origin or? |
| 23:26 | <dekiss> | Hixie ah ok man sry :) |
| 23:27 | <dekiss> | I never understood who make html and how it's made and this mailing system |
| 23:27 | <Hixie> | i'm not saying you can't ask here, just that if you want answers, this isn't hte best place for you :-) |
| 23:28 | <Hixie> | at least, if you want useful answers |
| 23:28 | <dekiss> | ok thanks |
| 23:28 | <dekiss> | man who makes HTML specification? |
| 23:28 | <dekiss> | you? ^^ |
| 23:28 | <dekiss> | editors? |
| 23:29 | <Hixie> | i write the text |
| 23:29 | <Hixie> | if that's what you mean |
| 23:32 | <ap> | Hixie: I honestly have a difficulty responding to <https://www.w3.org/Bugs/Public/show_bug.cgi?id=21416>;. There is a word "associate" in the algorithm that I don't understand. What is the expected result of doing an "associate"? |
| 23:33 | <ap> | Hixie: is it that both sides will immediately know that they are "associated"? |
| 23:33 | <dekiss> | I ment who is making the web these days? |
| 23:33 | <dekiss> | HTML standard |
| 23:33 | <ap> | Hixie: in such case, I just don't know how to implement that |
| 23:35 | <Hixie> | ap: is there any observable way to tell if it's instant or not? |
| 23:35 | <Hixie> | dekiss: if you mean http://whatwg.org/html, then i write the text, based on feedback here and on the mailing list, etc |
| 23:36 | <dekiss> | thanks that was what I ment |
| 23:36 | <dekiss> | Hixie one more quetion, will menu be disbanded? |
| 23:37 | <dekiss> | and rarely used elements? |
| 23:37 | <Hixie> | menu? |
| 23:37 | <ap> | Hixie: unsure - the string "associat" is not found elsewhere in text (other than as "association list", which I think is different). So the effect of "association" is implicit and thus difficult for me to verify |
| 23:38 | <dekiss> | yes "menu" element |
| 23:38 | <dekiss> | hm I am not sure I tihink I saw in w3c.org spec that it was not used much and is considered to be disbanded |
| 23:38 | <dekiss> | also hgroup |
| 23:38 | <Hixie> | ap: it means "let them be entangled" |
| 23:39 | <Hixie> | dekiss: <menu> and <hgroup> are new, i have no plans to remove them |
| 23:39 | <Hixie> | <hgroup> in particular is used in over 100,000,000 pages already |
| 23:39 | <ap> | Hixie: the spec defines "entangle" through "associate", so explaining "associate" through "entangle" doesn |
| 23:39 | <ap> | t |
| 23:39 | <ap> | help |
| 23:40 | <Hixie> | ap: elsewhere in the spec there are algorithms that refer to whether a port is entangled with another or not, and that refer to the entangled port. |
| 23:40 | <Hixie> | ap: this algorithm is attempting to create that connection. |
| 23:40 | <ap> | Hixie: or is "entagle" only called on ports that have not been shipped yet? |
| 23:41 | <zewt> | grr stumbling over new Event() because of the bad option defaults |
| 23:42 | <ap> | Hixie: so there is the clone algotithm: |
| 23:42 | <ap> | If the original port is entangled with another port, then run these substeps: |
| 23:42 | <Hixie> | ap: entangling of a port is something that happens upon its creation, and when its entangled port is cloned (when you clone a port A into a new port B, where A and C are entangled, A and C are disentangled and B and C are entangled) |
| 23:42 | <ap> | Let the remote port be the port with which the original port is entangled. |
| 23:43 | <ap> | Hixie: how can one detect if the original port is entangled? maybe another side has already initiated disentanglement, but it's still unknown to this side |
| 23:44 | <ap> | Hixie: e.g. ports A and B are entangled, and are in separate processes already. Scripts call clone() on them at the same time |
| 23:44 | <Hixie> | ap: you have to act as if it's instantaneous. Suppose you have A-B entangled, and you clone A to A' and B to B' at the "same time". You should end up with A'-B' entangled. |
| 23:45 | <ap> | Hixie: I doubt that there are implementations that do that |
| 23:45 | <ap> | Hixie: WebKit's certainly doesn't seems like it does |
| 23:46 | <Hixie> | ap: does webkit support cloning ports at all yet? |
| 23:48 | <ap> | Hixie: is this what you do when passing a port via postMessage? yes, that works (hopefully correctly when workers are not involved, and kind of maybe when they are) |
| 23:49 | <Hixie> | ap: well if it doesn't work i guess you have a bug :-) |
| 23:50 | <ap> | Hixie: it is of course something we can think of ourselves eventually, but it would be helpful if someone could think about whether the spec is implementable without unacceptable performance problems |
| 23:51 | <ap> | Hixie: such as pausing all processes to synchronize |
| 23:52 | <ap> | Hixie: anyway, I think that you answered bug 21416 - in that it's supposed to behave as if there was a cross-process mutex around all MessagePort operations |
| 23:52 | <Hixie> | ap: this doesn't require any synchronisation |
| 23:52 | <zewt> | drawing a blank for some reason: any examples of APIs that dispatch events, other than really weirdo things (in particular other than element.click) |
| 23:52 | <zewt> | er |
| 23:53 | <zewt> | that dispatch events synchronously |
| 23:53 | <ap> | zewt: execCommand :-p |
| 23:53 | <Hixie> | ap: just make sure you forward all your messages appropriately (and essentially leave a forwarding address) |
| 23:53 | <zewt> | ap: i'll bundle that in with weirdo things :P |
| 23:53 | <Hixie> | zewt: seach for "fire a " in the html spec |
| 23:54 | <zewt> | that'll have hundreds of hits, most of which are async |
| 23:54 | <Hixie> | oh synchronously |
| 23:54 | <zewt> | sorry, I mean JS entry points that fire events synchronously to script |
| 23:54 | <Hixie> | my bad |
| 23:54 | <Hixie> | dispatchEvent() is the main one |
| 23:55 | <ap> | Hixie: no, I understand that leaving a forwarding address is the next step. It's just that then I need to rip apart most of the other algorithms in the spec that don't explicitly account for that (most importantly those that explain object lifetime) |
| 23:55 | <zewt> | yeah (that's the one I was looking to compare to, actually) |
| 23:56 | <Hixie> | ap: like what? i'm happy to make the spec clearer, i just don't understand what the problem is here |
| 23:56 | <zewt> | (and I don't want to use click() as an example, there's so much horribly wrong with onclick it makes a bad comparison) |
| 23:57 | <ap> | Hixie: I _think_ that for every sentence that has the word "entangled" in it, I will need to think about whether it's actually "entangled or a forwarder to entangled" |
| 23:58 | <ap> | Hixie: and devise a protocol for dropping the forwarder privileges eventually |
| 23:58 | <Hixie> | ap: sure, that's like for every object that mentioned HTMLElement objects, you have to think about whether it means the C++-side of the object or the C++-side plus its JS-engine wrapper, or whatever is going on in the rendering engine. Figuring out these things is kind of your job. :-) |
| 23:58 | <ap> | Hixie: because when I have process A and process B, and a port gets shipped across the boundary and then back, we probably don't want to keep doing two cross-process messages any more |
| 23:59 | <Hixie> | ap: sure |
| 23:59 | <Hixie> | ap: doing cross-process IPC with pipes whose endpoints can be moved around the world is definitely non-trivial |
| 23:59 | <Hixie> | ap: i don't think anyone would argue otherwise :-) |
| 23:59 | <Hixie> | ap: our goal here is to try to make it as trivial as possible for Web authors, but that means eating a lot of pain |