2014-11-01 [17:54:12.0000] :( Safari doesn't have getElementById in the DocumentFragment prototype [18:19:19.0000] :[ [02:49:50.0000] Did anyone ever end up filing bugs on browsers for http://example.com/%F0%9F%94%92 and similar spoofing? [03:36:58.0000] It seems zcorpan filed a bug and it was a duplicate of https://bugzilla.mozilla.org/show_bug.cgi?id=808234 [09:43:54.0000] I filed https://github.com/github/github-services/issues/964 on github-services not dealing with URLs in commit messages properly, fwiw [10:00:16.0000] Domenic: http://jxck.hatenablog.com/entry/whatwg-stream [14:02:33.0000] Hixie: I can't subscribe to whatwg@whatwg with my new email address because the instructions in it reference lists.whatwg.org and the alternate method of replying to whatwg-requests⊙lwo gives an "undeliverable" [14:03:38.0000] annevk: wow, diagrams and everything!! [15:08:16.0000] mounir, /quit nn [15:10:58.0000] euhm, wtf rubys [15:23:03.0000] annevk: I think it's just a misunderstanding, albeit one that could spiral out of control... assume good faith and don't make things worse :) [15:23:14.0000] also dammit i keep emailing myself 2014-11-02 [17:16:20.0000] Domenic: which instructions? [17:34:27.0000] Hixie: when you email the subscribe address [17:35:06.0000] Hixie: basically, try to subscribe to whatwg and you'll see it's pretty bad [18:13:20.0000] ah [18:13:22.0000] yeah [18:13:24.0000] dunno what to do about it [18:14:42.0000] Domenic: wait, why is it bad [18:14:49.0000] Domenic: can't you just reply to the e-mail and be done with it? [18:16:15.0000] Hixie: those instructions don't work either. They contradict themselves. The reply to address is different from the one in the body and they tell you to have only that string. Try it. [18:16:48.0000] ok, i sent an e-mail ta whatwg-subscribe⊙wo [18:16:50.0000] i got back an e-mail [18:16:53.0000] i replied to it [18:17:00.0000] now i'm subscribed [18:17:03.0000] i don't understand the problem [18:17:12.0000] i mean, other than the e-mail mentions the website, which doesn't work [18:27:16.0000] Hmm replying didn't work for me [18:27:31.0000] And the reply-to didn't match the instructions in the body of the email [18:27:54.0000] Are there three separate ways to confirm subscription, only one of which works? [18:28:10.0000] Instructions in body for emailing whatwg-request don't work [18:28:18.0000] Clicking on the link doesn't work [18:28:24.0000] But replying does? [18:29:04.0000] there are three, two of which should work [18:29:18.0000] i haven't tried whatwg-request [18:31:36.0000] (the url will work too, you just have to use a browser that doesn't have HSTS information blocking that host) [18:32:19.0000] i sent a help command to whatwg-request⊙wo and that worked too, fwiw [18:53:00.0000] Trying to full understand the "is" attribute, but having a hard time searching for such a popular term. Anyone have a link to a good explanation and the spec? [18:53:23.0000] s/full/fully [18:54:33.0000] just had to ask so that i could find it. http://w3c.github.io/webcomponents/spec/custom/#instantiating-custom-elements [18:54:38.0000] never mind [18:56:37.0000] though if anyone has a good article, that's appreciated [22:14:51.0000] Hixie: hmm my bad, I typoed "request" :( [22:17:54.0000] heh [01:50:19.0000] @counter-style is part of css3, right? [02:09:31.0000] It's part of CSS, yes [02:19:02.0000] ok, Ms2ger, just wanted to make sure its correct if i tag it as CSS3 in MDN [02:19:43.0000] That's not particularly useful, but *shrug* [02:21:03.0000] alright.. i thought that might help if someone is searching for CSS3 specific things [02:22:13.0000] jsx: CSS3 is not a thing per the CSS WG [02:22:49.0000] oh! ok [02:23:02.0000] jsx: so if we pretend that CSS3 is a thing in MDN, we might want to overhaul that [02:23:15.0000] jsx: and instead focus on modules [02:23:19.0000] nope, i'm removing it rightaway [02:23:38.0000] thanks for the info [02:29:30.0000] annevk: is it a political thing like how the whatwg calls it html while the w3c calls it html5? or is this universal. (sorry, i'm very new to this) [02:30:55.0000] jsx, no, this is the CSSWG's call, actually [02:31:05.0000] jsx: while CSS 2.1 was still being developed, CSS3 (at the time still a thing) was developed in the form of modules [02:31:28.0000] jsx, they have version numbers for individual modules, but "CSS3" as a general thing doesn't exist [02:31:36.0000] jsx: however, some of those modules turned out to advance quicker than CSS3 as a whole [02:31:53.0000] jsx: so versioning (leveling in CSSWG terms) became scoped to modules, rather than CSS [02:32:23.0000] like "CSS Counter Styles Level 3" annevk? [02:32:29.0000] jsx: exactly [02:32:40.0000] jsx: http://www.xanthir.com/b4Ko0 has an explanation from TabAtkins [02:32:46.0000] TIL: Chrome still doesn't support setting event types to the empty string [02:32:53.0000] Speaking of which... [02:32:56.0000] annevk, r? https://github.com/w3c/web-platform-tests/pull/1342 :) [02:33:34.0000] thanks annevk, Ms2ger [02:33:41.0000] Np [02:33:59.0000] Ms2ger: I take it you ran these through a browser already? [02:34:55.0000] Er [02:35:11.0000] Looks like I broke one since the last time I did [02:36:23.0000] Ms2ger: createEvent table does not match https://dom.spec.whatwg.org/#dom-document-createevent [02:38:32.0000] Ms2ger: reviewed [02:38:40.0000] Thanks [02:38:49.0000] Looks like TouchEvent isn't supported in Fx? [02:38:59.0000] Oh, I guess it might be preffed off [02:40:38.0000] annevk, again? :) [02:40:49.0000] Ms2ger: again? [02:40:58.0000] I added the missing ones [02:41:34.0000] Ah, done [02:41:41.0000] Thanks [02:41:43.0000] I thought I had said something stupid somewhere [02:43:19.0000] No no :) [04:59:11.0000] Hixie: stylesheet on whatwg.org has "padding; 0;" [04:59:54.0000] Hixie: same stylesheet (tabbed-pages) also has -moz-outline which is dropped in Firefox Nightly [06:59:43.0000] The descriptions of those old DOM tests are pure poetry [06:59:51.0000] An document type accessed, setNodeValue is called with a non-null argument, but getNodeValue [06:59:52.0000] should still return null. [07:53:30.0000] annevk: That blog post gets so much mileage. [13:48:16.0000] TabAtkins: the things about URLs like css4-backgrounds could be updated, I guess [13:48:31.0000] though it might still be relevant for some TR snapshots 2014-11-03 [16:18:04.0000] looks like the only safe way of querying the contents of a DocumentFragment is to insert it into the document [16:23:33.0000] since /getElements?By(Id|TagName|Class)/ are over on Element, but the DocumentFragment interfact exposes only Node [00:01:01.0000] terinjokes: DocumentFragment should expose querySelector() [00:01:14.0000] terinjokes: and getElementById() [00:01:38.0000] terinjokes: if a browser doesn't expose those, bug [00:05:13.0000] annevk: Safari doesn't expose getElementById [00:05:45.0000] the other browsers I'm developing against pre-date querySelector and querySelectorAll… so can't blame them for not having it implemented [00:05:56.0000] sorry, Safari 8* [00:07:03.0000] http://msopentech.com/blog/2014/03/19/http2-nearing-completion/ HTTP WG makes decisions through coin toss... [00:14:55.0000] annevk, maybe htmlwg should try that [00:15:19.0000] annevk, I'm not sure their track record for good decisions beats a coin toss [00:16:34.0000] Ms2ger: if we don't count copying WHATWG work as decisions... [00:16:47.0000] I don't think we do [01:21:33.0000] Domenic: how would we ever expose Fetch' terminate primitive? [01:21:55.0000] Domenic: var res = await fetch(req); req.terminate()? [03:49:55.0000] how does one see when https://github.com/w3c/web-platform-tests/blob/master/workers/interfaces/DedicatedWorkerGlobalScope/postMessage/event-ports-dedicated.html was added originally to wpt? [03:50:04.0000] github UI isn't good with file moves [03:50:07.0000] apparently [03:50:13.0000] not with huge changes [03:55:33.0000] (that test seems to be invalid) [04:16:14.0000] smaug____: it's prolly added by zcorpan [04:18:10.0000] zcorpan: why e.ports === null ? [04:19:06.0000] The spec seems to say "Let new ports be an empty array." ... "Let the ports attribute of the event be initialised to the new ports array." [04:21:13.0000] smaug____: probably it matched the spec when the test was written. my interpretation of the current spec is that it should be empty array [04:21:45.0000] smaug____: https://github.com/w3c/web-platform-tests/commit/eb29e8c9fcc0e8a8e685581123bd1380de207d7d is maybe when it was first added [04:23:06.0000] zcorpan: http://www.w3.org/TR/webmessaging/#messageport requires empty array [04:23:12.0000] that is from 2012 [04:23:23.0000] MikeSmith: don't see that file in that commit. but i wrote the test in question [04:23:53.0000] smaug____: i think the test predated Transferable thing [04:23:56.0000] ah [04:24:30.0000] but it's also possible that i misread the spec of course [04:25:53.0000] smaug____: do you want the spec changed or hte test changed? [04:27:11.0000] zcorpan: OK, but anyway seems like it is actually there and https://github.com/w3c/web-platform-tests/blob/eb29e8c9fcc0e8a8e685581123bd1380de207d7d/Workers/tests/approved/interfaces/DedicatedWorkerGlobalScope/postMessage/event-ports-dedicated.html is the original (unless I'm missing something) [04:41:01.0000] zcorpan: test [04:41:29.0000] or I don't care too much which on should be changed [04:41:33.0000] one [04:48:29.0000] http://lists.w3.org/Archives/Public/public-w3process/2014Nov/0015.html chaals arguing to disable TLS to preserve battery... You cannot make this shit up [04:51:00.0000] wow [06:05:13.0000] smaug____: do you have a github username? [06:06:36.0000] smaug____: https://github.com/w3c/web-platform-tests/pull/1347 [06:47:40.0000] zcorpan, good idea re: interfaces.html [06:47:49.0000] Didn't think it would work, but it does [07:38:02.0000] http://dnscurve.org/ seems interesting [08:51:37.0000] TabAtkins: let me know when I should try again. Meanwhile, I'd also appreciate a pointer to how to produce biblio entries. [08:52:38.0000] rubys: https://github.com/tabatkins/bikeshed/blob/master/docs/definitions-autolinks.md#bibliography [08:52:46.0000] And what do you mean by "try again"? [08:53:39.0000] TabAtkins: "Ugh, this sent my comment before it was done. One sec while I finish up..." [08:53:47.0000] Oh, it's there now. [08:53:55.0000] ok, cool [08:55:34.0000] rubys: Are you trying to maintain all the current "concept-foo-bar"-style urls, or are you planning to switch them all into real autolinks? [08:56:49.0000] TabAtkins: I am willing to do whatever. The question isn't what I'm willing to do, but what annevk is willing to accept. [08:56:56.0000] Haha, ok. [08:57:09.0000] rubys: I'm a bit confused with the IDL issue [08:57:45.0000] rubys: e.g. as I understand it https://w3c.github.io/webappsec/specs/content-security-policy/#securitypolicyviolationevent-interface is done with bikeshed and has links [08:58:03.0000] rubys: does that mean Mike accepted the maintenance cost? [08:58:05.0000] annevk: All Bikeshed-generated. [08:58:29.0000] You can put your s outside of the IDL block, if you want, and Bikeshed will link to them from the IDL (instead of trying to dfn them in the IDL). [08:58:32.0000] Anolis requires a fair amount of markup to make IDL pretty... I don't really mind it [08:59:01.0000] annevk: Bikeshed requires *more* markup to make the IDL work correctly, if you're doing it by hand. [08:59:14.0000] You have to specify the types and the "for" values for everything. [08:59:27.0000] TabAtkins: okay, so you are saying Mike is not doing it by hand [08:59:28.0000] TabAtkins: that's what I was missing. I should be able to work with that. [08:59:59.0000] so... generate, see what dfns are produced, move them to where they belong, and... done. [09:00:02.0000] rubys: Ah, sorry, didn't realize there was a disconnect there. Just make sure your external dfns are marked up right, with types and "for" values. [09:00:21.0000] TabAtkins: how do you link [Constructor] accurately without markup? [09:00:34.0000] Just like foo(int arg) [09:00:51.0000] sweet [09:00:52.0000] annevk: I think I do it badly right now. It's on my list of fixes to make. [09:01:23.0000] TabAtkins: okay, so the only thing that's not supported is that in interface URL, URL becomes the ? [09:01:34.0000] No, that works too. [09:01:38.0000] oh [09:01:52.0000] Okay so rubys maybe misunderstood something then? [09:01:54.0000] /me is so confused [09:01:57.0000] There was a bug for a little bit in some related handling, but I fixed that this morning. [09:02:25.0000] Yeah, I think rubys didn't realize that Bikeshed'll respect you wanting to put your s elsewhere. [09:02:28.0000] annevk: I think that about summarizes it. Everything works, I just didn't know how. Perhaps there is a doc problem, or perhaps it is just me. [09:02:45.0000] Well, it should all be in the docs, but docs are hard to read, so wtv. [09:03:49.0000] rubys: while you're here, please check with me next time if you think I say something weird [09:04:37.0000] annevk: you say lots of things that are weird :-) [09:04:53.0000] rubys: sure, but those two www-tag emails were quite upsetting [09:05:52.0000] Sorry. I will say that I'm still confused by your assertion that reverse engineering existing browses is incompatible with what the WHATWG does. [09:06:20.0000] rubys: I haven't said that [09:07:12.0000] rubys: I've said that you can't document reality since reality is that browsers disagree, whereas WHATWG writes documents that dictates them to agree [09:08:03.0000] reality is more complicated than that. There are cases where the browsers agree, and what you have speced is different than that. [09:08:23.0000] rubys: for those cases it might make sense to change the specification [09:09:12.0000] in which case, I will claim that your statement on this matter is confusing. [09:10:13.0000] For those wanting to follow along, we are referring to: http://lists.w3.org/Archives/Public/public-whatwg-archive/2014Nov/0004.html [09:10:14.0000] rubys: either I've made a mistake in analyzing the data when I wrote the specification, there's more data now, or the critique is inaccurate or from insufficient data [09:10:17.0000] What I said: [09:10:25.0000] > That may not be as we would wish it to be. But it would be a disservice to > everyone to document how we would wish things to be rather than how they > actually are (and, by all indications, are likely to remain for the > foreseeable future). [09:10:28.0000] What anne said: [09:10:34.0000] This contradicts with most WHATWG work. WHATWG standards describe how things should be, taking into account the realities of deployed content. [09:11:14.0000] I'd like to continue trying to find the sweet spot between what you have spec'ed, what browsers are willing to accept, and how far I can push the IETF. [09:11:34.0000] may i suggest that instead of arguing over the metaissue of what you meant when arguing what the goals are, you instead just focus on the issues that matter, namely, the actual bugs? [09:11:41.0000] I don't understand what's unclear. Your statements contains absolutes that are just untrue. [09:12:05.0000] Hixie: good advice. [09:12:43.0000] Hixie: first step on that road is here: http://intertwingly.net/projects/pegurl/urltest-results/ [09:12:57.0000] Yeah, I've been missing that. There's a lot of statements "from above", but no bug reports yet. [09:13:25.0000] rubys: file a bug with what you want changed, providing the evidence you think supports that change [09:13:25.0000] Which is a weird state to be in given that rubys wants it to be non-political. [09:14:00.0000] annevk: unfair. I have submitted pull requests. I'm trying to work first on items that you have identified (bikeshed and rewriting the parser part of the spec). I think that's a fair amount of work. [09:14:21.0000] Hixie: you might also find the following to be of interest: http://intertwingly.net/projects/pegurl/liveview.html [09:14:55.0000] since i'm not editor of the url spec and have interest in getting involved in the url spec... :-) [09:15:47.0000] Hixie: I'm having difficultly parsing that. Perhaps s/have/don't/? [09:15:56.0000] rubys: I appreciate that work, quite a lot. But it's no basis for changing direction relative to RFC 3986, for instance. [09:15:58.0000] have no, rather, yeah [09:16:10.0000] Hixie: ack [09:17:09.0000] annevk: I'm going to keep doing the work. I am also going to try to work with everybody. That includes yourself, the TAG, the IETF, browser vendors, library implementors, etc. [09:17:45.0000] rubys: Sure, I do the same thing [09:18:14.0000] If you say something that is different than something Domenic says, my preference would be to get the two of you chatting. [09:18:48.0000] I think Domenic and I are quite well aligned [09:19:10.0000] Not sure what you're referring to [09:19:13.0000] I've now done that twice. The first time you agreed to accept pull requests. The second time you found what I had done to be quite upsetting. [09:21:44.0000] Yes, as I explained, your statement contained absolutes that are untrue. The URL Standard documents how we wish things to be given that there are browsers out there that do not conform to it. It cannot document how things are, given that browsers disagree. Perhaps there are cases where those two could be closer, but you've yet to file a bug for those. [09:21:52.0000] Annevk: a more recent example where you confused me: you suggested "hiding the editor field through CSS". When I pointed out how it could be done, you said " It seems wrong to me to have it there and use display:none to hide it." [09:22:42.0000] rubys: your selective quoting does not help, I said "The other thing I don't like, but won't block on, is hiding the editor field through CSS." [09:23:09.0000] that's fair, and I agree with it; but I will still say that I'm confused as to what your position is. [09:23:14.0000] rubys: I knew from TabAtkins that hiding the editor field through CSS was the way to go, but I don't like it [09:23:43.0000] rubys: and in that comment I indicated that I wouldn't block on converting the URL Standard to bikeshed because of that [09:24:12.0000] From my experience with other open source efforts; I feel that removing editors entirely (and not just moving it to the bottom as you have done), is the right way to go. [09:25:40.0000] Seems reasonable [09:26:01.0000] rubys: might be worth pointing out here https://github.com/tabatkins/bikeshed/issues/255#issuecomment-61509373 [09:26:03.0000] The reason I'm currently in favor of leaving the editors in (and hiding them through CSS if you dont' want to display them) is that there are existing biblio-scraper tools that know how to look for Editors formatted in the way I currently generate them, and build bibliography dbs from scraping a bunch of docs. [09:26:21.0000] annevk: will do. [09:26:23.0000] TabAtkins: rubys seems to argue that it would just say WHATWG [09:26:33.0000] annevk: +1 [09:26:48.0000] If you really want to indicate no editors, I have a bug open on me to allow you to indicate you're omitting Editor on purpose. [09:27:16.0000] Sure, that's easy to do too. "Editor: WHATWG, http://whatwg.org, whatwg⊙wo" [09:27:38.0000] TabAtkins: WFM [09:27:48.0000] TabAtkins: rather no editors, with org being WHATWG [09:28:14.0000] Like we do with Unicode references [09:28:46.0000] For context: the Apache Software Foundation has a strong position on there being no technical leads for ANY of its products. [09:29:32.0000] I'm not quite sure about going that far [09:29:45.0000] that's the exact opposite of how the WHATWG works :-) [09:29:53.0000] You can even put that Editor line into the defaults-whatwg.include file, so all whatwg specs get it automatically. [09:30:11.0000] every spec here has one person to blame [09:30:12.0000] by design [09:30:16.0000] so we know how to blame [09:30:21.0000] (Yeah, I'm actually quite unclear on where the editor-minimization is coming from, given that whatwg generally BDFLs...) [09:30:40.0000] the idea of moving the credit for editing from the header to the footer is just one of modesty [09:30:50.0000] since the credit should go to all the people sending feedback [09:31:11.0000] In this case, the problem is that the modesty interferes with the boilerplate generation in a very annoying way. [09:31:14.0000] the whatwg model is one of concentrating the responsibility and blame, and distributing the credit [09:31:23.0000] there shouldn't be any boilerplate :-) [09:31:30.0000] (e.g. the html spec doesn't have any) [09:31:42.0000] My sense is that Dominic and Marcos are more open to sharing the responsibility. [09:31:52.0000] sharing responsibility is a bug [09:32:06.0000] it leads to sharing the blame [09:32:14.0000] it is imho one of the main causes of the w3c's specs being low quality [09:32:18.0000] /me goes back to working on real bugs [09:32:35.0000] Hixie: You keep saying that, and you keep being wrong. ^_^ But that's because HTML is a special snowflake, and nobody is trying to be that consistent with its style. [09:32:54.0000] wait, what was it i said that was wrong? [09:32:59.0000] TabAtkins, ... style consistency? [09:33:00.0000] /me doubts that he could ever mod Bikeshed enough to satisfy HTML's use-cases. [09:33:00.0000] i said a lot of things just then :-P [09:33:17.0000] Hixie: The thing you were replying to me about - boilerplate. [09:33:32.0000] oh i didn't realise i'd ever said that html had no boilerplate before [09:33:36.0000] it's a relatively new fact [09:33:38.0000] You typed a bunch more between when I started typing and when I was done, sorry. [09:33:41.0000] as of ~august [09:33:43.0000] Speaking of boilerplate [09:33:56.0000] The multipage button has the old url [09:34:08.0000] When I was first working with Domenic to build the WHATWG boilerplate, you kept claiming there shoudlnt' be any, and everyone should build craft their header content by hand. [09:34:09.0000] Hixie, ^ [09:34:37.0000] TabAtkins: ah, yeah, i definitely stand by that. if you have much boilerplate, you're probably doing it wrong, imho. [09:34:43.0000] Ms2ger: interesting [09:35:00.0000] Ms2ger: oh, yeah, that's intentional [09:35:05.0000] Ms2ger: it's still shorter than the new one [09:35:07.0000] There's a difference between "too much" and "more than you want to write by hand, and duplicate-with-minor-changes into new specs". [09:35:15.0000] That's true [09:35:24.0000] TabAtkins: *shrug* [09:35:48.0000] TabAtkins: the html spec really has virtually nothing that you'd duplicate into other specs [09:35:56.0000] TabAtkins: so i stand by what i said [09:36:06.0000] TabAtkins: i feel that boilerplate is stuff that turns readers away [09:36:12.0000] The HTML spec is a special snowflake :) [09:36:17.0000] if you don't want to write it, why would anyone want to read it [09:36:30.0000] all our specs should be special snowflakes [09:36:32.0000] I think the header for https://dom.spec.whatwg.org/ makes sense [09:36:36.0000] otherwise why are we writing them? [09:36:51.0000] And I don't see why other specs would need a different style [09:37:47.0000] anyway [09:37:51.0000] y'all do as you wish :-) [09:38:05.0000] And w'all will be consistent ;) [09:38:26.0000] Hixie: it's hard to reconcile your statement for consistent look & feel with desire for inconsistent headers [09:38:37.0000] Hixie: I dont' want to write all the markup for my IDL blocks either, but people seem to get value from that. [09:38:47.0000] annevk: which statement for consistent look & feel? [09:38:58.0000] Hixie: sharing stylesheet [09:39:24.0000] annevk: if there's a good reason for a spec to have a different look, then it shouldn't share the style sheet either [09:41:44.0000] annevk: done. https://github.com/tabatkins/bikeshed/issues/255#issuecomment-61516495 [09:45:27.0000] As for the discussion about BDFL... It has been useful for someone to make the decisions and be responsible for addressing issues. I agree that if there's no such feeling issues tend to go unaddressed or get weak responses from committees. [09:46:58.0000] If we removed editor information even from the Acknowledgments I would still expect someone to be responsible, even if not written down somewhere. [09:48:46.0000] Python works well with a BDFL. Apache projects work effectively WITHOUT a BDFL. We can all cite efforts that don't work well. Some with BDFLs, and many without. [09:49:50.0000] rubys: http://www.apache.org/foundation/ suggests there's some oversight at least [09:50:14.0000] rubys: with impressive VP titles even [09:50:31.0000] Indeed. There is oversight. But VPs are not technical leads. By design. [09:50:53.0000] I've been on the Apache Board of Directors for many years. If you have any questions, let me know. [09:51:18.0000] I've been release lead (something we rotate), PMC chair, and many other positions too. [09:51:39.0000] In a way editors are not technical leads either. That's the various browsers [09:52:48.0000] annevk: at the risk of upsetting you again, I see little evidence of that in http://intertwingly.net/projects/pegurl/urltest-results/ [09:53:43.0000] ok, let me retract "little evidence". That was overstating the problems I'm seeing. [09:55:30.0000] rubys: it's obvious you are seeing something in those test results I'm missing out on [09:56:34.0000] annevk: scroll to the bottom of http://lists.w3.org/Archives/Public/public-webapps/2014OctDec/0315.html for three examples [09:57:45.0000] rubys: example 1) http://intertwingly.net/projects/pegurl/urltest-results/7357a04b5b all browsers seem to agree on serializing except for IE [09:57:51.0000] rubys: why would you side with IE? [09:58:04.0000] annevk: look at username and password [09:58:09.0000] rubys: example 2) http://intertwingly.net/projects/pegurl/urltest-results/4b60e32190 we discussed that the results are different on Windows [09:58:21.0000] rubys: I don't think all browsers implement username and password properties [09:58:33.0000] rubys: that's a new feature [09:58:50.0000] annevk: the spec doesn't differentiate between "running on windows" and "running elsewhere". [09:59:07.0000] rubys: yes, and I explained why [09:59:26.0000] rubys: if you feel we should go for platform-dependent parsing, okay... [09:59:44.0000] no. I think we should remove what one browser implemented on one platform. [09:59:51.0000] http://intertwingly.net/projects/pegurl/urltest-results/61a4a14209 I already conceded we need to revisit IPv4 parsing [10:00:00.0000] and look what presto did for: http://intertwingly.net/projects/pegurl/urltest-results/7357a04b5b [10:00:07.0000] Presto is irrelevant [10:00:15.0000] I think a strong case can be made for hiding passwords [10:00:18.0000] They gave up their browser and its URL parser because it broke too many sites [10:01:27.0000] rubys: hiding doesn't help if you can still get it through getAttribute() or some such [10:02:11.0000] annevk: this is a topic that merits more discussion. With more browser vendors participating. I met with several last week. [10:03:28.0000] rubys: I don't think URL's username/password options make much sense, but they're not really harmful either, and are used by XMLHttpRequest's HTTP authentication options [10:04:00.0000] rubys: if anything we allow disabling them at the parser level [10:04:06.0000] we should allow* [10:04:23.0000] WFM [10:06:54.0000] Anyway, I already said IP parsing needed looking into. File URLs can use another careful look if someone has the time. [10:07:00.0000] I don't really see any controversy about that [10:08:02.0000] And I don't see how browsers doing wildly different things for them contradicts about what I said that we'd need to make a decision on what they should do instead. Since just matching reality is obviously not going to work... [10:10:31.0000] you seem stuck on that. I said something along the lines of determining what they were willing to do, and you said that contradicts how the WHATWG works. I suggest that we both move on from that discussion, and focus on working with everybody and finding the sweet spots. [10:11:29.0000] that sounds like a good thing [10:12:56.0000] rubys: determining what browsers are willing to do is not what you said, but seems like a good thing, yes [10:13:37.0000] cool. peace. onwards! [10:19:04.0000] Seems hz-gb-2312 is dead https://www.w3.org/Bugs/Public/show_bug.cgi?id=25339 [10:19:09.0000] Party time? [10:19:15.0000] JakeA: annevk: should it be possible to put a Response with type "error" in the cache? [10:20:16.0000] wanderview: don't really see a reason why not, although it also seems rather pointless [10:20:44.0000] annevk: if fetch() returns an "error" response... it rejects its promise, right? [10:20:57.0000] wanderview: yeah, TypeError [10:21:23.0000] wanderview: agree with annevk [10:21:23.0000] not NetworkError? [10:21:29.0000] wanderview: no [10:21:53.0000] wanderview: didn't want to create a dependency on DOM for the sake of Node.js [10:22:53.0000] Removing hz-gb-2312 is exciting people [10:23:12.0000] i'm excited [10:23:15.0000] JakeA: is it just me, or does addAll() not say what to do if one of the requests fails? [10:27:56.0000] JakeA: nm... this explains it for me: https://github.com/slightlyoff/ServiceWorker/issues/541 [10:30:21.0000] wanderview: oh, it should be reject with the first error (like Promise.all) [10:34:42.0000] JakeA: I think thats what it says... (promise verbage is new to me) [10:37:05.0000] JakeA: do you have an opinion on this? https://github.com/slightlyoff/ServiceWorker/issues/549 [10:40:18.0000] JakeA: I'm sorry but I forgot timeline updates, when can we expect updates to the spec & issues again? [10:44:38.0000] annevk: from me? [10:44:51.0000] JakeA: yeah and others I guess that work on the spec [10:47:24.0000] annevk: I'm flying back today so normal service starts resuming (although I made some minor fixes to cache.match a couple of days ago). Looks like Jungkee had been pretty constantly active? [10:48:45.0000] Is there a particular issue you have in mind? I've still got request.client on my todo list, but wanted to do that when we're on same time zone [10:49:14.0000] JakeA: 445 [10:49:22.0000] JakeA: also the path issue [10:49:34.0000] JakeA: where everyone missed out on the security hole [10:49:56.0000] JakeA: and other issues that impact implementations, such as ServiceWorkerClient design [10:50:19.0000] JakeA: environment settings object stuff; creating and terminating workers [10:50:33.0000] JakeA: message channel setup for service workers [10:57:46.0000] http://www.bis.doc.gov/index.php/component/content/article/107-about-bis/newsroom/press-releases/press-release-2014/763-intel-subsidiary-agrees-to-750-000-penalty-for-unauthorized-encryption-exports :-( [11:02:27.0000] annevk: which path issue did everyone miss out on? [11:03:35.0000] (did some searching, couldn't find anything, but I'm on a throttled roaming connection so it's like wading through mud) [11:05:05.0000] JakeA: https://github.com/slightlyoff/ServiceWorker/issues/468 [11:05:15.0000] JakeA: also explained in 445 [11:06:40.0000] annevk: you mean the cache poisoning thing? I wouldn't say no one thought of that, I mention it a couple of comments before you did https://github.com/slightlyoff/ServiceWorker/issues/468#issuecomment-60352628 :P [11:07:18.0000] Will reply now [11:07:34.0000] JakeA: oops, missed that [11:08:28.0000] JakeA: seems like a bad thing to have this security theatre then [11:11:04.0000] annevk: we've protected existing badly structured sites. But if they remain badly structured and start relying on origin storage (even just IDB), they're opening themselves up to attacks from others on the origin [11:12:32.0000] But they have to open the door by wrongly assuming IDB/localstorage/cache is secure to them [11:22:57.0000] Some naming help needed. Bikeshed currently categorizes spec versions into two categories - "ED" and "TR" - and assumes the ED is more up-to-date. This is confusing when applied to specs outside the W3C. Suggestions on (non-derogatory) names for the categories? [11:23:04.0000] Maybe "latest" and "snapshot". [11:24:23.0000] current and stale [11:24:34.0000] Oh, I missed the parenthetical [11:24:48.0000] "current" and "snapshot" might work. [11:33:51.0000] TabAtkins: git would call them branch & tag [11:34:40.0000] Dunno if that's helpful. Tbh "current" and "snapshot" sounds better [11:34:42.0000] Yeah, but then you're invoking very specific git idioms (and "branch" isn't even quite right; maybe "master") [11:36:03.0000] TabAtkins: although w3 html is definitely a branch at this point [11:36:14.0000] That's even worse! ^_^ [11:36:29.0000] No, a fork :) [11:36:35.0000] Hah [11:37:07.0000] Fork and branch are identical in the wider git data model. [11:44:24.0000] annevk: I've got next week to spend on issues. Maybe Wednesday afternoon too, but will be just off a 23hr flight so who knows. Thursday morning too. Will chat to you about priorities. [11:47:47.0000] TabAtkins: Thanks for the (non-derogatory) parenthetical. I prefer "latest" and "current". Give me a few minutes to see if I can come up with something better than "snapshot". [11:53:23.0000] TabAtkins: I think "stable" is better than "snapshot". "dated" works too, if you don't mind being mildly subversive. [11:54:36.0000] Dated sounds good [11:54:47.0000] "current' and "dated" it is. [11:54:48.0000] And then Fig'd for current? [11:55:14.0000] "Fig'd"? [11:55:24.0000] Date <-> Fig [11:55:34.0000] BOOOOOOO [11:55:39.0000] :D [12:08:34.0000] TabAtkins: consider adding "numbered" and "names" as synonyms for "dated". [12:08:52.0000] "names"? [12:09:16.0000] named* [12:09:40.0000] None of those have the pun, though! [12:10:50.0000] Up to you, just tossing ideas out there. [12:39:52.0000] JakeA: ta [13:46:40.0000] Ms2ger: you're terrible and you should feel bad [13:47:02.0000] :( [13:47:11.0000] jwalden would have approved [13:47:21.0000] \o/ [13:47:24.0000] wait what [13:47:30.0000] /me consults logs [13:49:59.0000] mm, figs [13:50:36.0000]