| 09:29 | <zcorpan> | why was sec-websocket-location dropped? |
| 09:32 | <jgraham> | Not sure. The only reference I can find is abarth saying: "This was a remnant from a previous draft that made use of that header. |
| 09:32 | <jgraham> | " |
| 09:32 | <zcorpan> | yeah |
| 09:32 | <jgraham> | (it got left in the IANA section for a bit) |
| 09:33 | <hsivonen> | what's the implentation status of sec-origin for cross-origin requests? |
| 09:36 | <hsivonen> | zcorpan: so the weird string after "Opera Mobi/" is a build id according to http://my.opera.com/community/openweb/idopera/ |
| 09:37 | <hsivonen> | that's a lot of bytes to send on every HTTP request to every site for info that only Opera Software ASA has useful use for |
| 09:38 | <hsivonen> | What's the purpose of the X-EBO-UA header that Opera Mobile sends? |
| 09:38 | <hsivonen> | googling for it doesn't help |
| 09:38 | <jgraham> | hsivonen: The explaination in the first point of https://developer.mozilla.org/en/QA/Avoiding_intermittent_oranges seems bogus |
| 09:40 | <hsivonen> | jgraham: yeah. the whole example was bogus |
| 09:41 | <hsivonen> | jgraham: when the example got fixed, the explanation stayed |
| 09:41 | hsivonen | fixes |
| 09:45 | <hsivonen> | jgraham: thanks. the explanation should be less bogus now. (may require reloading. the site has more-agressive-than-average caching policies) |
| 09:46 | <jgraham> | hsivonen: That looks better, thanks :) |
| 09:56 | <zcorpan> | hsivonen: ok |
| 14:00 | <zcorpan> | websocket review http://www.ietf.org/mail-archive/web/hybi/current/msg07063.html |
| 14:34 | <zcorpan> | MikeSmith: everything ok with publishing? |
| 14:34 | <MikeSmith> | zcorpan: yeah |
| 14:36 | <zcorpan> | great |
| 14:36 | <zcorpan> | lemme know if i need to do something more |
| 14:44 | <hsivonen> | gotta love Roy's take on Gerv's statement on the licensing question |
| 14:48 | <jgraham> | I'm only surprised he didn't try to claim that the solution to the licensing problem was in his thesis |
| 16:10 | <TabAtkins> | AryehGregor: You still need an opinion on the <center> email? |
| 16:20 | <TabAtkins> | AryehGregor: If so (since there's no relevant followups on the thread) you are correct - there is no way to replicate <center> with currently-implemented CSS. <center> is doable with Flexbox, but not with a single inline style on the element itself - you have to set margins to 'auto' on the direct children. This is similar to, but not as restrictive as, the current practice of centering with auto margins, since flexbox items can be cen |
| 16:21 | <jgraham> | TabAtkins: "since flexbox items can be cen…" |
| 16:21 | <TabAtkins> | The <center> behavior is thus specified as "center { text-align: center; display: flexbox; flex-direction: block; } center > * { margin-left: auto; margin-right: auto; }" |
| 16:21 | <TabAtkins> | "centered without having a specified width." |
| 17:28 | <Deltachaos> | hi @all |
| 17:29 | <Deltachaos> | i got a idea. there are lot of services using certificates out there. what about to have certificates in browsers for auth to? |
| 17:31 | <MikeSmith> | Deltachaos: like WebID? |
| 17:31 | <MikeSmith> | http://www.w3.org/wiki/WebID |
| 17:31 | <Deltachaos> | so you can set your private and public key in the browser and if you visit a website witch wants you to autherificate your browser shows up a bar witch looks like the geolocation. |
| 17:34 | <Deltachaos> | MikeSmith: oh i dont know about that. i look :D |
| 17:36 | <MikeSmith> | Deltachaos: not sure, but I think it's sorta the same thing you just described |
| 17:36 | <MikeSmith> | there's a related mailing list too |
| 17:36 | <MikeSmith> | http://lists.w3.org/Archives/Public/public-xg-webid/ |
| 17:36 | <MikeSmith> | might be useful to look through the archives of that |
| 17:36 | <MikeSmith> | and/or subscribe to it |
| 17:36 | <MikeSmith> | it's a public list that anyone can subscribe to and post messages to |
| 17:36 | <Deltachaos> | well on the first look it looks interresting. but is it used for autherification? |
| 17:42 | <MikeSmith> | Deltachaos: you mean authentication? |
| 17:42 | <Deltachaos> | yes :D sorry for my english^^ im from germany |
| 17:42 | <MikeSmith> | OK |
| 17:42 | <MikeSmith> | well, I know very little about the details of WebID |
| 17:42 | <MikeSmith> | but looking at the wiki, I think it is meant as a means for providing authentication |
| 17:43 | <MikeSmith> | [[ |
| 17:43 | <MikeSmith> | The value of having a URI as an identifier are: |
| 17:43 | <MikeSmith> | that it can be linked to by other profiles, to create a linked web of trust |
| 17:43 | <MikeSmith> | that it can be tied to information enabling a method of authentication ( such as OpenID or even more directly with foaf+ssl ) |
| 17:43 | <MikeSmith> | ]] |
| 17:43 | <MikeSmith> | (quoting the wiki) |
| 17:44 | <MikeSmith> | [[ |
| 17:44 | <MikeSmith> | The WebID Protocol authenticates a digital identity (a WebID) by allowing an Agent (e.g., a Web Browser) to prove possession of or access to a private key, whose corresponding public key is tightly bound to that WebID. The private key is usually associated with a "certificate" on the user's computer, while the public key and WebID are part of that certificate and tightly bound to its subject. Likewise, the public key and WebID are |
| 17:44 | <MikeSmith> | tightly bound into the profile-oriented document(s) that describe the identity being authenticated. |
| 17:44 | <MikeSmith> | ]] |
| 17:44 | <MikeSmith> | so, it seems like it's a kind of authentication at least |
| 17:44 | <wilhelm> | A web of trust or a hierarchy of blame? |
| 17:44 | <MikeSmith> | heh |
| 17:45 | MikeSmith | wonders what foaf+ssl is |
| 17:46 | <MikeSmith> | has OpenID had much success so far? |
| 17:46 | <MikeSmith> | it seems like I don't hear about it so much these days |
| 17:48 | <MikeSmith> | Deltachaos: anyway, maybe you can write up a message describing what specific use case / problem case you have in mind, and asking whether WebID addresses that case or not |
| 17:48 | <MikeSmith> | to the public-xg-webid⊙wo list, I mean |
| 17:49 | <Deltachaos> | ok |
| 19:02 | <AryehGregor> | TabAtkins, thanks. |
| 19:14 | AryehGregor | ponders deleteContents() |
| 19:15 | <AryehGregor> | I guess the spec is fairly sane, all things considered. |
| 19:27 | <AryehGregor> | It's just unnecessarily complicated . . . |
| 19:27 | <AryehGregor> | No, hmm, I don't think it's sane. |
| 19:27 | <AryehGregor> | Because it behaves differently for the case where a node is partially selected and the case where it's not. |
| 19:32 | <AryehGregor> | Oh, no, it doesn't really. |
| 19:32 | <AryehGregor> | So it is sane, basically. Just pointlessly complicated. |
| 19:59 | <MikeSmith> | AryehGregor: deleteContents() is pretty old stuff, isn't it? |
| 19:59 | <AryehGregor> | MikeSmith, what do you mean? |
| 20:00 | <MikeSmith> | I mean isn't it already implemented? |
| 20:01 | <AryehGregor> | Yes. |
| 20:02 | <AryehGregor> | But not fully specced. |
| 20:02 | <MikeSmith> | ah |
| 20:04 | <AryehGregor> | Only vaguely specced. |
| 20:04 | <MikeSmith> | so it can't be changed at this point, but at least can be properly specced still |
| 20:04 | <AryehGregor> | Well, it could be changed in minor ways if that made it more logical. |
| 20:04 | <MikeSmith> | what spec is it actually in? |
| 20:04 | <MikeSmith> | ok |
| 20:04 | <AryehGregor> | But I don't think there's a good case for it. |
| 20:04 | <MikeSmith> | I see |
| 20:04 | <AryehGregor> | The old spec is DOM 2 Range: http://www.w3.org/TR/DOM-Level-2-Traversal-Range/ranges.html |
| 20:04 | <AryehGregor> | The new spec is here: http://html5.org/specs/dom-range.html |
| 20:04 | <MikeSmith> | ah |
| 20:04 | <MikeSmith> | wow, the original one is vintage 2000 |
| 20:04 | <AryehGregor> | Yeah. |
| 20:04 | <AryehGregor> | Actually not too terrible, as such specs go. |
| 20:04 | <AryehGregor> | Much more precise than any CSS spec I've read. |
| 20:04 | <AryehGregor> | There are relatively few points whose interpretation is debatable, even for corner cases. |
| 20:04 | <AryehGregor> | Although there are a handful of points where it's wrong (does not match implementations). |
| 20:04 | <MikeSmith> | ok |
| 20:04 | <MikeSmith> | are you working on the current DOM Range spec also? |
| 20:04 | <AryehGregor> | Yes. |
| 20:04 | <AryehGregor> | ("also" meaning "in addition to execCommand()" or what?) |
| 20:04 | <Ms2ger> | In addition to being awesome in general? :) |
| 20:04 | <MikeSmith> | I mean in addition to Ms2ger |
| 20:04 | <AryehGregor> | Oh, yes. |
| 20:04 | <Ms2ger> | More than me :) |
| 20:04 | <AryehGregor> | In the last few months, more than him, yeah. |
| 20:05 | <MikeSmith> | if you all have been putting a lot of time into it, I guess I should actually take time to read it |
| 20:06 | <Ms2ger> | I guess I should, too |
| 20:06 | <MikeSmith> | heh |
| 20:06 | <MikeSmith> | does it add any new interfaces, or is it currently just documenting existing implementation behavior? |
| 20:06 | <Ms2ger> | Just documenting |
| 20:06 | <MikeSmith> | ok |
| 20:07 | <MikeSmith> | does it have testing impact? |
| 20:07 | <Ms2ger> | It has excellent tests |
| 20:07 | <MikeSmith> | that is, will it need a new test suite? |
| 20:07 | <MikeSmith> | ok |
| 20:07 | <Ms2ger> | And I can say that because Aryeh wrote them all |
| 20:07 | <MikeSmith> | ah, sweet |
| 20:08 | <MikeSmith> | great to see that somebody is on top of things as far as testing |
| 20:08 | <AryehGregor> | http://aryeh.name/spec/dom-range/test/ |
| 20:08 | <AryehGregor> | Ian made writing tests part of my contract terms, which I was happy with, since I'm all for it. |
| 20:08 | <MikeSmith> | cool |
| 20:09 | <MikeSmith> | are the tests automated in some way? or automatable if plugged into the right harness? |
| 20:09 | <AryehGregor> | They use jgraham's harness, so totally automatic. |
| 20:09 | <MikeSmith> | excellent |
| 20:09 | <AryehGregor> | Just visit the page and it gives you the results. Plus you can hook it up to something that will run them all in the background and tally everything up. |
| 20:09 | <AryehGregor> | It's easy because they're pure JS. |
| 20:09 | <PrgmrBill> | I think TDD is awesome and I write tests voluntarily :P |
| 20:09 | <PrgmrBill> | I'm trying to set an example where I work (I'm lead developer) |
| 20:09 | <AryehGregor> | TDD being what? |
| 20:09 | <PrgmrBill> | Test Driven Development |
| 20:09 | <Ms2ger> | PrgmrBill, want to write some for HTML? :) |
| 20:10 | <MikeSmith> | AryehGregor: btw, I like how your .swp file shows up in that directory listing |
| 20:10 | <PrgmrBill> | haha I think tests should be written by the developer |
| 20:10 | <AryehGregor> | MikeSmith, yeah, I'm not aware of another place where the dom-range tests are actually runnable. |
| 20:10 | <MikeSmith> | ok |
| 20:10 | <AryehGregor> | bitbucket lets you download them, but not run them. |
| 20:11 | AryehGregor | is in the middle of fiddling with some of them, namely deleteContents, so that one's likely to not work so well |
| 20:11 | <Ms2ger> | http://ms2ger.freehostia.com/tests/runner/?path=../dom-range/ |
| 20:12 | <Ms2ger> | Though I should update them |
| 20:12 | <MikeSmith> | PrgmrBill: that approach works less well when you have one person writing the spec, who is essentially the architect, and you have multiple developers writing separate implementations |
| 20:12 | <AryehGregor> | MikeSmith, it works well for reverse-engineered specs. |
| 20:12 | <AryehGregor> | The spec author can just write the tests. |
| 20:13 | <MikeSmith> | so maybe we can get Hixie to do that for all the reverse-engineered features in the HTML spec :) |
| 20:13 | <Ms2ger> | Also doesn't work when the editor has spec writing work left for a few decades |
| 20:16 | <MikeSmith> | I was going to suggest we need a cloneHixie() method, but I'm suddenly reminded of The Sorcerer's Apprentice |
| 20:16 | <bfrohs> | Question: best way to mark up a three-column form? (Field Name | Input | Description of field and/or messages about errors while filling it out) |
| 20:17 | <Ms2ger> | We certainly do |
| 20:17 | <AryehGregor> | Hurrah, I just found another place where DOM 2 Range doesn't quite match reality. |
| 20:29 | <MikeSmith> | Nth earthquake of the month just now |
| 20:30 | <MikeSmith> | becoming variation on the "how often" joke? visitor to Tokyo: "How often do earthquakes happen in Tokyo?" response: "So often that you'll just quit noticing them." |
| 20:30 | <MikeSmith> | …except you don't actually ever quit noticing them |
| 20:51 | <AryehGregor> | What are PIs actually used for in practice, in XML? |
| 20:51 | <AryehGregor> | (Or is the answer "they're not"?) |
| 20:51 | <AryehGregor> | <?xml-stylesheet ?> ? |
| 20:52 | <Ms2ger> | Yup |
| 20:54 | <AryehGregor> | Should we a) get ProcessIngInstruction changed to implement CharacterData, or b) not allow PIs to be range endpoints? Because the status quo is just stupid. |
| 20:54 | <AryehGregor> | Way too many spec lines are wasted on PIs. |
| 20:54 | <AryehGregor> | IE and Opera already don't let them be range endpoints. |
| 20:55 | <AryehGregor> | Spec lines and spec-writing hours. |
| 20:55 | AryehGregor | solicits Ms2ger's opinion |
| 20:55 | <AryehGregor> | I think (b) is easier. |
| 20:55 | <AryehGregor> | We just have to update all the places where we throw exceptions for DocumentType to also throw for ProcessingInstruction. |
| 20:56 | <Ms2ger> | I can change PIs in DOM Core, but that means all implementations have to change, for little gain |
| 20:56 | <AryehGregor> | http://www.w3.org/Bugs/Public/show_bug.cgi?id=12205 |
| 20:56 | <Ms2ger> | I was that :) |
| 20:56 | <Ms2ger> | saw, even |
| 20:57 | <AryehGregor> | Anne and Olli both think this makes sense, but Alexey and Adrian have doubts. |
| 20:57 | <AryehGregor> | So how about I just change it so PIs can't be endpoints? Are you okay with that? |
| 20:57 | <Ms2ger> | Fine with me |
| 20:57 | <AryehGregor> | Okay, will do. |
| 21:00 | <MikeSmith> | I think PIs were intended to be used for non-standard things, and I seem to remember that people argued against using a PI as standard means for binding a stylesheet to a document |
| 21:02 | AryehGregor | discovers that "length" was never actually defined for DocumentTypes, and no one noticed |
| 21:04 | <AryehGregor> | The only place it seems to matter is where I defined "before" and "after". |
| 21:08 | <AryehGregor> | Or something. |
| 21:50 | <MikeSmith> | I have a long HTML <ul> list of links that I'd like to copy into a mediawiki wiki |
| 21:50 | <MikeSmith> | can I import it somehow? |
| 21:50 | MikeSmith | finds http://toolserver.org/~diberri/cgi-bin/html2wiki/index.cgi |
| 22:03 | <MikeSmith> | I create a list of specs that would benefit from having a common testing framework |
| 22:03 | <MikeSmith> | http://www.w3.org/wiki/Testing/Specs |
| 22:04 | <MikeSmith> | anybody/everybody please feel free to add to it |
| 22:04 | <MikeSmith> | it currently doesn't list any CSS specs at all |
| 22:04 | <MikeSmith> | but it should |
| 22:04 | <AryehGregor> | Okay, pushed the PI endpoint change. Probably breaks lots of stuff. |
| 22:04 | <AryehGregor> | MikeSmith, isn't that "everything"? I mean, what specs would not benefit from a common testing framework? |
| 22:05 | <MikeSmith> | specs that aren't intended to be implemented in browsers |
| 22:05 | <MikeSmith> | or that aren't implementable in browsers |
| 22:06 | <AryehGregor> | Well, okay. |
| 22:06 | <AryehGregor> | So why don't you call the list "specs relevant to browsers"? |
| 22:10 | <MikeSmith> | yeah, could call it that too |
| 22:10 | <MikeSmith> | I also just made one for listing relevant groups: |
| 22:10 | <MikeSmith> | http://www.w3.org/wiki/Testing/Groups |
| 22:10 | <MikeSmith> | which is even less complete |
| 22:12 | <MikeSmith> | I guess that list of specs is useful in other ways than just in the context of testing |
| 22:25 | <TabAtkins> | AryehGregor: I'm thinking of chopping the Hebrew list style to just cover the range 0-999, based on http://www.smontagu.org/writings/HebrewNumbers.html complaining about the ambiguity and no widely-established method. |
| 22:25 | <TabAtkins> | Does that sound roughly reasonable? |
| 22:26 | <TabAtkins> | (You're my go-to Hebrew writer.) |
| 22:26 | <AryehGregor> | TabAtkins, it's not uncommon for book numbers or such to go over 1000, but they usually just follow the same pattern, e.g., תתת for 1200. |
| 22:26 | <AryehGregor> | The pattern with the ` in it isn't so widely established in my experience. I've only seen it very rarely. |
| 22:27 | <AryehGregor> | For instance, the convention for year numbers is just to omit the thousands part, so this year is תשעא = 771, although actually it's 5771. |
| 22:27 | <AryehGregor> | So I wouldn't stop it at 999, that's arbitrary. |
| 22:28 | <AryehGregor> | You want to stop it at some point so it doesn't get totally ridiculous, because asymptotically the length of the written number is linear in the size of the number. |
| 22:28 | <AryehGregor> | IIRC, though, Gecko implements the spec as written for numbers over 1000. You'll want to check what implementations do before changing the spec. |
| 22:28 | <AryehGregor> | But all else being equal, I'd say it's safer to just extend the pattern and not special-case numbers above 999. |
| 22:29 | <TabAtkins> | Ah, hm. That's easy to do, actually, by just turning Hebrew back into an additive system. Then it'll just repeat TAV until it gets small enough to fill in the rest of the value, up to whatever range I specify. |
| 22:29 | <AryehGregor> | Yeah, that would make sense. |
| 22:29 | <AryehGregor> | But as usual on Hebrew issues, you probably want to be asking Israelis, not American Jews, since Israelis are going to be the primary users of Hebrew stuff. |
| 22:29 | <AryehGregor> | And perspectives will differ. |
| 22:30 | <TabAtkins> | Or I just leave the range out, and let excessively-large numbers be covered by my general paragraph about implementation limits on the various counter styles. |
| 22:30 | <TabAtkins> | Given that symbolic and additive both have the possibility for an author to really blow out a list marker. |
| 22:30 | <TabAtkins> | Hell, just define unary and then set @value to something huge. |
| 22:31 | <TabAtkins> | @counter-style unary { type: additive; additive-glyphs: 1 "1"; } |
| 22:37 | <TabAtkins> | Anyway, I think I'll ping Aharon and see what the Tel Aviv office thinks. |
| 22:42 | <AryehGregor> | jgraham, have you considered using something other than tables for testharness output? Because formatting a 20,000-line table is really, really slow. |
| 22:43 | <AryehGregor> | Maybe we could just use divs with percentage-width sub-divs? |