| 00:10 | <JonathanNeal> | Domenic_: min assumes a zero-like value? |
| 00:11 | <JonathanNeal> | Whoops, very delayed response to your question. |
| 00:28 | <JonathanNeal> | The really tragic part about all of this is that JAWS 15 removed support for <hgroup> where it had previously supported it. |
| 00:28 | <JonathanNeal> | So, we actually took a step back. |
| 00:29 | <JonathanNeal> | I'm sorry, I meant to say that it supported the outline algorithm (sectioning elements). Not sure about <hgroup> though I imagine they were all bundled together. |
| 00:33 | <zcorpan> | annevk: including .float is actually intended. but there's a use counter for dashed properties and for float, so we'll see what that gives |
| 00:33 | <Domenic_> | JonathanNeal: why do care what the W3C spec says? |
| 00:33 | <jamesr__> | semantically correct website: http://motherfuckingwebsite.com/ |
| 00:33 | <jamesr__> | actually, can someone check if his use of <strong> is appropriate? |
| 00:34 | <jamesr__> | i think it's right |
| 00:34 | <Domenic_> | Haha speaking of subheadings he uses <aside> for that |
| 00:35 | <jamesr__> | Note: It's not appropriate to use the aside element just for parentheticals, since those are part of the main flow of the document. |
| 00:35 | <jamesr__> | is that a parenthetical, or is it separable from the main flow of the document? |
| 00:35 | <Domenic_> | It seems parenthetical. Certainly not separable. |
| 00:35 | <jamesr__> | i think it's sufficiently separable to use an <aside> |
| 00:36 | <Domenic_> | ok i just realized by uttering those words i made myself into a monster and so am leaving this conversation now. |
| 00:36 | <jamesr__> | i think we have to fight now |
| 00:37 | <jamesr__> | or at least file formal objections against each other |
| 00:38 | <JonathanNeal> | I think Domenic_ is suggesting a <PAREN> tag. |
| 00:38 | <JonathanNeal> | </SARCASM> |
| 00:39 | <zcorpan> | Domenic_: the blockquote usage is wrong (per whatwg spec at least) |
| 00:39 | <JonathanNeal> | That's true, plus you should use <footer> for cites. |
| 00:40 | <zcorpan> | not according to the spec |
| 00:40 | <Domenic_> | zcorpan: cuz he's missing <p>s? |
| 00:40 | <Domenic_> | http://developers.whatwg.org/grouping-content.html#the-blockquote-element |
| 00:40 | <zcorpan> | Domenic_: no |
| 00:41 | <Domenic_> | oh because the person's name is inside the quote |
| 00:41 | <zcorpan> | yes |
| 00:41 | <zcorpan> | and the quote marks should probably be omitted also |
| 00:41 | <zcorpan> | of course some people like hsivonen think that that usage is correct and the spec is wrong |
| 00:42 | <JonathanNeal> | I care because I write websites with subheadings all the time, and it frustrates me when a bunch of us create websites and guides to writing websites that rely on some new feature, and then it gets removed, nullifying a bunch of advice, offering no equivalent solution, and later noting a lack of adoption. |
| 00:42 | <JonathanNeal> | To respond to your initial question, Domenic_. |
| 00:43 | <Domenic_> | JonathanNeal: no, that doesn't really answer the question. My question was more, "why do you think the W3C spec is any more relevant than King Lear is?" |
| 00:44 | <JonathanNeal> | Hey, no need to randomly insult the works of Shakespeare. |
| 00:44 | <JonathanNeal> | I think it's relevant because it obviously has an impact on implementation and adoption. |
| 00:45 | <Domenic_> | Really? |
| 00:46 | <JonathanNeal> | Yes. To quote from earlier "the problem with <hgroup> is that the FUD that the w3c spec has added to the mix means that it'll be even longer than normal for this to get implemented widely, unfortunately". |
| 00:46 | <Domenic_> | Heh, fair. |
| 00:47 | <Domenic_> | I guess, my approach is to not let the W3C spec affect how I author or teach. I can see having to deal with the fallout from its existence though. |
| 00:48 | <JonathanNeal> | Interesting to see the difference in the solutions, <figure><blockquote/><figcaption/></figure> BS <blockquote><p>[...]<footer/></blockquote>. |
| 00:49 | <JonathanNeal> | In regards to what you just shared earlier. I never knew there was alignment issue with blockquote recommendations too. |
| 00:50 | <Domenic_> | the number of alignment issues is numerous and growing |
| 00:52 | <JonathanNeal> | And scrolling down http://html5doctor.com/blockquote-q-cite/#comment-23768 reveals that the post may be inaccurate. |
| 00:53 | <Domenic_> | the thing i love about the html5doctor comments section is the number of hidden tags from people who forgot they had to type < and > |
| 00:54 | <JonathanNeal> | I still agree with http://www.w3.org/html/wg/wiki/ChangeProposals/hgroup which is why I rally this subject every so often. |
| 00:55 | <JonathanNeal> | But <hgroup> is closer to what is needed, way better than nothing. |
| 00:55 | zcorpan | finds http://lists.w3.org/Archives/Public/public-html/2011Nov/0031.html |
| 01:35 | <SteveF> | JonathanNeal: "<JonathanNeal> The really tragic part about all of this is that JAWS 15 removed support for <hgroup> where it had previously supported it. " JAWS never supported hgroup or proper outline nesting |
| 01:36 | <SteveF> | Domenic_:"<Domenic_> JonathanNeal: why do care what the W3C spec says? you may not care but lost of dev's do... |
| 01:37 | <JonathanNeal> | SteveF: I made a correction to that quote. |
| 01:38 | <JonathanNeal> | And they did support outline nesting, moreso then than now. |
| 01:39 | <SteveF> | Jonatahan:Neal: yeah saw that after, unfortunately it was borked, we tried to work with them to fix it but they were not interested |
| 01:39 | <SteveF> | thier implementation of the oultine was borked that is |
| 01:42 | <JonathanNeal> | Help me understand what I perceive to be a dichotomy. You're contributing to a spec that you're also telling people is dangerous fiction to follow, in regards to sectioning elements. |
| 01:45 | <JonathanNeal> | And then, why not remove the sectioning part of elements from the w3 draft if they're dangerous, as <hgroup> was removed since it relied so heavily on them. |
| 01:45 | <SteveF> | JonathaNeal: no dichotomy, I have modified the advice/requirements in the HTML spec to provide guidance on the use of hx, the outline algorithm is still there sure maybe one day it will become useful and then the advice heading use can be changed, either way. |
| 01:47 | <JonathanNeal> | What is the difference between removing a feature and "sure maybe one day"-keeping it? |
| 01:48 | <SteveF> | JonathonNeal: there are no implementation requirements on browsers for the outline algorithm |
| 01:50 | <SteveF> | JonathanNeal: but sure if the WG decide that the outline algorithm is a waste of space then it will get dropped, we haven't reached that point yet |
| 01:54 | <SteveF> | JonathanNeal: if you think there needs to be clearer advice re fiction of outline then please file a bug or email the html list or me or whatever. |
| 01:55 | <JonathanNeal> | It was my hope that dropping <hgroup> meant getting something better. Your <subline/subhead> draft did that, but it never made it into a spec. Now there's a hole (an acknowledged hole) in the mutually agreed upon outlining algorithm. I've followed http://www.w3.org/html/wg/wiki/ChangeProposals/hgroup for years now. I'm sad to see it stale. |
| 01:55 | <SteveF> | JonathanNeal: then unstale it |
| 01:56 | <SteveF> | JonathanNeal: things change if people do stuff |
| 01:58 | <JonathanNeal> | Yea, I'm working on the great barrier of logging into the system. |
| 02:00 | <JonathanNeal> | I'd like to harmonize the current git work with the w3 wiki, which might ultimately help get the w3 spec and the whatwg spec in more harmony, but I can't use my w3.org creds on the w3 wiki. |
| 02:01 | <SteveF> | Domenic_: also note that the differences in the 2 specs are almost all about author conformance requirements, so browser implementer folk need not care if they don't wish to, conformance checker implementers do (in the case of main for example, which was developed at w3c and rewritten differently by hixie) and that is reflected in implementations. |
| 02:02 | <Domenic_> | As an author, I pay attention to the source material, not a fork. |
| 02:03 | <SteveF> | Domenic_: fair play to you, many devs don't take your view |
| 02:03 | <Domenic_> | I think most devs listen to w3schools :P |
| 02:04 | <SteveF> | Domenic_:yeah and they dropped hgroup a while back... |
| 02:04 | <Domenic_> | haha |
| 04:42 | <MikeSmith> | gsnedders: that validator message you saw isn't because of that server not resolving. It's expected that it doesn't resolve. The validator uses an internal resolver to map those URLs to locally cached copies of the schemas. |
| 09:00 | <hsivonen> | annevk-cloud: https://bugzilla.mozilla.org/show_bug.cgi?id=910187#c2 surprises me considering that we stopped recognizing the Armscii-8 label in Firefox 19 and no other browser engine support it |
| 09:01 | <Ms2ger> | Hmm, Armscii-8 |
| 09:01 | <Ms2ger> | I think I still have a patch lying around to remove the decoder |
| 09:23 | Ms2ger | wonders where MikeSmith is on http://www.w3.org/2013/11/w3cteam |
| 09:25 | <jgraham> | Ms2ger: Pretty sure I would deny knowing MikeSmith too ;) |
| 09:25 | <SteveF> | Ms2ger: front far right |
| 09:26 | <SteveF> | he looks photoshopped |
| 09:26 | <Ms2ger> | Heh |
| 09:27 | <jgraham> | The whole photo looks very strange |
| 09:27 | <jgraham> | I think it had a lot of noise reduction applied to it or something |
| 09:30 | <darobin> | jgraham: I don't think that much noise reduction was applied to it |
| 09:30 | <darobin> | I mean I saw the original, it wasn't very noisy |
| 09:34 | <Ms2ger> | Oh, darobin is on there too? |
| 09:34 | <darobin> | damn right |
| 09:35 | <Ms2ger> | "Nothing precludes the author from using custom elements in HEAD." |
| 09:35 | <darobin> | Ms2ger: I'm roughly in the middle, grey tee with sunglasses hanging from it |
| 09:35 | <darobin> | yeah, that made my day too |
| 09:36 | <Ms2ger> | With the weird grin? |
| 09:36 | <jgraham> | darobin: Well there must be some reason everything looks so smudgy, except at areas of high contrast where there is a kind of halo effect |
| 09:37 | <darobin> | jgraham: I think that's because it may have been compressed a bit too much |
| 09:37 | <jgraham> | Ms2ger: Which message is that from? |
| 09:38 | <darobin> | Ms2ger: possibly :) |
| 09:38 | <Ms2ger> | jgraham, http://lists.w3.org/Archives/Public/public-webapps/2013OctDec/0610.html |
| 09:38 | <darobin> | jgraham: webapps |
| 09:38 | <jgraham> | darobin: Hmm, I guess that might well have a similar effect |
| 09:38 | <jgraham> | darobin: Right, webapps isn't a message :p |
| 09:38 | <Ms2ger> | darobin, also, is that a burrito you're holding? |
| 09:38 | <darobin> | yeah yeah, there were like five messages overnight |
| 09:39 | <darobin> | Ms2ger: well spotted! but no |
| 09:39 | <jgraham> | Ah. Well good to know that the people in charge of our glorious future understand how the platform works |
| 09:39 | <darobin> | it's a pinny as they say over the Channel |
| 09:40 | <darobin> | we made dumpings |
| 09:40 | <Ms2ger> | jgraham, it does open body, right? |
| 09:41 | <darobin> | yes |
| 09:42 | <darobin> | but hey, nothing indeed prevents you from putting whatever you want between <head> and </head> :) |
| 09:42 | <jgraham> | http://software.hixie.ch/utilities/js/live-dom-viewer/saved/2665 |
| 09:43 | <jgraham> | I mean nothing does strop you putting stuff in <head> |
| 09:43 | <jgraham> | As long as you don't actually use the HTML parser |
| 09:43 | <darobin> | right |
| 09:43 | <jgraham> | For example you could send your document in JSON and insert all the content using the DOM |
| 09:43 | <jgraham> | Actually |
| 09:44 | <jgraham> | Pretend I didn't say that |
| 09:44 | <jgraham> | Otherwise someone will think it's a great idea |
| 09:44 | darobin | trots off to make a spec for it |
| 09:44 | <darobin> | I'll put it in the same document in which I define a z attribute on every SVG element so that we have declarative 3D |
| 09:50 | <Ms2ger> | jgraham, surely it would be more readable to use the html5lib tests tree format |
| 10:11 | <Ms2ger> | Hixie, thanks for the acid3 update |
| 10:43 | <MikeSmith> | I photoshopped myself into that photo, as SteveF correctly surmised. I photoshopped in darobin too, which is why he's grinning. |
| 10:43 | <MikeSmith> | At the the time that photo was taken I was away for a bit helping the Shenzhen Internet Police refine some of their man-in-the-middle systems. |
| 10:45 | <Ms2ger> | I see |
| 10:45 | <Ms2ger> | That explains why you aren't listed in the description |
| 10:53 | <darobin> | and I had fled the country to escape the slashback for w3cmemes |
| 10:54 | <MikeSmith> | heh |
| 11:01 | <MikeSmith> | I heard w3cmemes barely made it out of China alive. |
| 11:01 | <MikeSmith> | Alternatively I heard that w3cmemes was seen crossing the border back to HK in the back seat of a PRC limousine, smoking a cigar. |
| 11:03 | <MikeSmith> | Ms2ger: I'm listed as "Bert Bos", which is an anagram for "Michael[tm] Smith" |
| 11:03 | MikeSmith | wonders what update Hixie made to Acid3 |
| 11:05 | <hsivonen> | annevk-cloud: check out the source on http://main.am/armenianchurch/Program/General/FS2-1.htm |
| 11:06 | <hsivonen> | that site is a fascinating combination of pre-Unicode and post-Acid3 Web tech |
| 11:07 | <hsivonen> | anyway, the report about armscii-8 support being needed lost its credibility, since that site doesn't actually use armscii-8 support |
| 11:16 | <Ms2ger> | MikeSmith, http://lists.w3.org/Archives/Public/www-archive/2013Nov/0013.html |
| 11:16 | <Ms2ger> | MikeSmith, and I've seen Bert Bos, sorry :) |
| 11:18 | MikeSmith | notes that Ms2ger says he's sorry that he's seen Bert Bos |
| 11:19 | <MikeSmith> | Ms2ger: thanks for the link |
| 11:19 | <annevk> | hsivonen: charset=x-user-defined o_O |
| 11:20 | <annevk> | oh lol, character references all the way down |
| 11:20 | <hsivonen> | annevk: right. it doesn't actually use the x-user-defined range |
| 11:20 | <hsivonen> | annevk: instead, it assigns the Armenian glyphs to the Latin-1 Supplement range |
| 11:20 | <hsivonen> | and it supplies a font via @font-face |
| 11:21 | <hsivonen> | so x-user-defined only has an effect if you have that same font installed locally in a pre-@font-face Firefox |
| 11:21 | <annevk> | wow, I thought that was an India only thing |
| 11:21 | <hsivonen> | annevk: me, too |
| 11:22 | <hsivonen> | anyway, not evidence of armscii-8 use |
| 11:22 | <hsivonen> | and explains how Chrome can succeed in Armenia |
| 11:25 | <MikeSmith> | hsivonen: you mean you think that there are a lot of sites doing that? |
| 11:25 | <MikeSmith> | (doing what that page is doing I mean) |
| 11:28 | <hsivonen> | MikeSmith: not sure what I should think. the author of that page claims there are more like that, but he/she already made a untrue claim, so I'm not sure what to believe |
| 11:29 | <MikeSmith> | I guess it's more likely that hack is probably not something he dreamed up himself |
| 11:30 | <MikeSmith> | but instead just copied from some other more widely used site |
| 11:31 | <hsivonen> | MikeSmith: that sort of commandeering Latin 1 characters for non-Latin glyphs is not an isolated incident |
| 11:31 | <hsivonen> | MikeSmith: used to be a thing in India |
| 11:32 | <hsivonen> | MikeSmith: dunno how common the exact combination of x-user-defined with entities is, though |
| 11:32 | <MikeSmith> | ah OK |
| 11:32 | <gsnedders> | MikeSmith: Then it shouldn't cause an exception trying to start! |
| 11:32 | <MikeSmith> | gsnedders: indeed |
| 11:33 | <hsivonen> | MikeSmith: anyway, what that site does today is interoperably supported |
| 11:33 | <hsivonen> | (armscii-8 isn't and wasn't) |
| 11:34 | <MikeSmith> | hsivonen: yeah the guy who posted that bug just seemed to not even understand what he was doing himself, or what was actually needed and being used in practice for Armenian |
| 11:35 | <MikeSmith> | (or the guy who commented there in the recent comment -- I guess he's not the one who raised the bug) |
| 11:35 | <MikeSmith> | gsnedders: did you make any changes the validator sources locally? |
| 11:35 | <gsnedders> | MikeSmith: No. |
| 11:36 | <gsnedders> | MikeSmith: I literally just followed the published instructions, checked out build than ran the Python script twice |
| 11:36 | <MikeSmith> | ok |
| 11:36 | <MikeSmith> | can you please try running it one more time? |
| 11:36 | <gsnedders> | Sure. |
| 11:37 | <hsivonen> | how come Emoji landed in Unicode before https://en.wikipedia.org/wiki/Arevakhach ? |
| 11:37 | <gsnedders> | (I don't actually care. I just want a small bit of the Validator.nu source. :P) |
| 11:37 | <gsnedders> | MikeSmith: Third time lucky, seems to be working |
| 11:37 | <MikeSmith> | gsnedders: I care if I broke something :) |
| 11:37 | <MikeSmith> | gsnedders: yeah I guess we need to modify the build instructions then |
| 11:38 | <MikeSmith> | "Re-run the build script until it doesn't fail any more." |
| 11:38 | <annevk> | in response to last night, seems markup mu has not yet been achieved |
| 11:38 | <hsivonen> | gsnedders: it's possible that I broke the appearance of something that was broken not being broken, if this is about it actually trying to fetch something from s.validator.nu |
| 11:38 | <gsnedders> | MikeSmith: :) |
| 11:38 | <gsnedders> | gsnedders: Yes |
| 11:38 | <gsnedders> | gsnedders: Why are you talking to yourself? |
| 11:38 | <gsnedders> | hsivonen: Yes. |
| 11:39 | <hsivonen> | gsnedders: that suggests our mapping of magic bogo URLs to local data is broken |
| 11:39 | <gsnedders> | hsivonen: But only on second run of build |
| 11:39 | <hsivonen> | weird |
| 11:54 | <MikeSmith> | hsivonen, gsnedders I think it's most likely something I messed up in the last few months. I'll try to look at it when I can make the time |
| 12:02 | <annevk> | Custom elements made it to CR? |
| 12:06 | <MikeSmith> | annevk: I thought it just went to LC recenlty |
| 12:06 | <annevk> | MikeSmith: I think the way it works is that nobody has even noticed that happened |
| 12:06 | <MikeSmith> | heh |
| 12:07 | <annevk> | MikeSmith: apart from rniwa from Apple it seems, but his alternative proposal was not seen as a comment |
| 12:09 | <MikeSmith> | annevk: I'm glad at least there's a plan in place to clear up the confusion in the future http://w3cmemes.tumblr.com/post/66838310443/disappointed-in-the-guardians-of-the-process |
| 12:11 | <MikeSmith> | annevk: if they missed an alternative proposal I doubt it was intentional |
| 12:11 | <MikeSmith> | I'll ask Yves |
| 12:12 | <annevk> | MikeSmith: it seems to be mentioned in http://www.w3.org/wiki/Webapps/LCWD-custom-elements-20131024/ so I guess I was missing something |
| 12:12 | <annevk> | MikeSmith: still not convinced that with four comments, all of the monkey patching going on, this is actually ready |
| 12:15 | <MikeSmith> | annevk: I think if Apple's saying they're not planning on implementing as specified it's not ready |
| 12:16 | <MikeSmith> | especially if they are saying they'll implement it if changes are made |
| 12:20 | <MikeSmith> | http://www.w3.org/wiki/Webapps/LCWD-custom-elements-20131024/#LC-2 seems to be dglazkov saying he plans to resolve the comment without making any changes |
| 14:07 | <annevk> | hsivonen: it seems Chrome does not respect x-user-defined as a label |
| 14:07 | <annevk> | hsivonen: at least not in HTML context |
| 14:50 | <darobin> | jgraham or Ms2ger: am I guessing correctly that all the WPT tests starting with html5lib_ in html/syntax/parsing are generated automatically from the html5lib test suite? |
| 14:51 | <darobin> | and if so, I guess that if I wanted to modify them it'd be best to patch upstream? which in turns begs the question of finding the converter and all |
| 14:51 | <Ms2ger> | Yes |
| 14:51 | <jgraham> | darobin: Yes |
| 14:51 | <Ms2ger> | The converter should be in wpt |
| 14:52 | <Ms2ger> | Under tools/? |
| 14:52 | <jgraham> | The converter is in tools |
| 14:52 | <darobin> | ah |
| 14:52 | darobin | digs |
| 14:52 | <Ms2ger> | And https://github.com/html5lib/html5lib-tests |
| 14:52 | <darobin> | found it, thanks a lot! |
| 16:48 | <zcorpan> | jgraham: if i want to test that ping=... got resolved correctly in the POST request, how should i do that? a naive way would be to stash the result and then do an XHR, but i think that has some problems like what if the XHR happens before the ping happens, what if ping isn't supported... |
| 16:50 | <zcorpan> | i guess i could feature-check for ping and fail early. if the feature check passes but the ping isn't sent anyway, the test can time out |
| 16:50 | <jgraham> | zcorpan: I guess you need to do the ping-causing action and then have a timeout to do the XHR |
| 16:50 | <jgraham> | If it can be delayed for an arbitary amount of time it seems impossible to tell for sure whether if succeeded or not |
| 16:51 | <zcorpan> | can't i do the XHR immediately and have the server script wait with returning anything until the stash is there? |
| 16:51 | <jgraham> | Well sure, I guess |
| 16:51 | <jgraham> | But that still needs some kind of timeout |
| 16:52 | <zcorpan> | isn't the harness timeout enough? |
| 16:53 | <jgraham> | On the server side I mean |
| 16:53 | <jgraham> | You can't juet let the script run forever |
| 16:54 | <zcorpan> | the script isn't killed if the test is navigated? |
| 16:54 | <jgraham> | No, how would that work? |
| 16:55 | <zcorpan> | the browser closes the XHR connection on navigation, i thought. but maybe it doesn't? |
| 16:55 | <jgraham> | Right, but the server won't notice that |
| 16:56 | <zcorpan> | ok. is that intentional? |
| 16:56 | <jgraham> | Well intentional in the sense that I don't know how to do it another way |
| 16:56 | <jgraham> | You would need to know whether the remote end had closed the socket. But I don't know how to tell without trying to send something over it |
| 16:57 | <zcorpan> | i was imagining some magic where the server would notice that the connection was closed and kill the script |
| 16:57 | <jgraham> | There isn't any such magic |
| 16:57 | <zcorpan> | k |
| 16:57 | <jgraham> | If someone knows how to add such magic, it might be a good idea |
| 16:58 | <zcorpan> | i don't know what the socket api is like. but it seems a bit weird to not be notified about the socket closing. but i recall pywebsocket also not knowing about it |
| 16:59 | <zcorpan> | maybe my mental model of how sockets work is just wrong |
| 17:11 | <zewt> | not completely following the conversation, but you can definitely tell if the other side has closed a TCP connection |
| 17:12 | <jgraham> | zewt: Using getsockopt? Or something else? |
| 17:13 | <zewt> | just by trying to read it; you'll get an EOF (and a read signal from select) |
| 17:20 | <jgraham> | Can I actually use that in this situation though? |
| 17:22 | <zewt> | not familiar with what you're doing ... are you testing against a dummy server to see what the browser is doing? |
| 17:23 | <jgraham> | I don't know what the difference between a dummy server and a real server |
| 17:23 | <jgraham> | is |
| 17:23 | <Ms2ger> | The words "dummy" and "real" |
| 17:23 | <jgraham> | But yeah, I have a server |
| 17:23 | <jgraham> | and it does some work |
| 17:24 | <jgraham> | and maybe, by the time the work is done, the client has closed the connection |
| 17:24 | <zewt> | a real server being apache or nginx or something that you're just making requests against, as opposed to something designed to inspect some particular behavior |
| 17:24 | <Ms2ger> | dglazkov, :) |
| 17:24 | <jgraham> | It's a real server in the sense that you don't directly access the inner state of the server |
| 17:24 | <jgraham> | when running tests |
| 17:29 | <zewt> | well, the server can tell if the client has closed the connection; whether you can make use of that in whatever it is you're doing with whatever server you're using I couldn't tell you :) |
| 18:32 | <SimonSapin> | annevk: https://code.google.com/p/chromium/issues/detail?id=324251#c2 says data: URLs have no fragment (it’s part of the "content") because rfc2397 doesn’t talk about it. I disagree. What do you think? |
| 18:33 | <Ms2ger> | What does the fo^Wspec say? |
| 18:33 | <annevk> | SimonSapin: you're correct, that's been proven long ago |
| 18:33 | <annevk> | SimonSapin: fragment identifiers are part of all URLs |
| 18:34 | <annevk> | SimonSapin: the way the IETF defines URIs with ABNF on a per-scheme basis makes this confusing admittedly; but the URL Standard solves that |
| 18:36 | <annevk> | SimonSapin: pretty sure that bug is a duplicate btw |
| 18:46 | <zcorpan> | jgraham: another thing i want to do is get the raw characters for a part of the request url. what i came up with was re.match(r'q=([^&]+)', request.url_parts.query).groups()[0] |
| 18:47 | <jgraham> | zcorpan: That seems like it would work |
| 18:49 | <jgraham> | I admit it isn't super-pretty, but in most cases request.GET will be good enough |
| 18:49 | <zcorpan> | yeah it seems to do the trick. at least when the url only has ascii-range bytes, iirc (old?) IE can request urls without percent-escaping and i don't know what would happen |
| 18:50 | <jgraham> | So making cases where it isn't good enough pretty seems like a lower priority |
| 18:50 | <zcorpan> | yeah i don't disagree |
| 18:52 | <jgraham> | You can go even lower level if something on the python side isn't playing well with non-ascii there e.g. request.url or request.request_line |
| 18:52 | <jgraham> | I think the regexp should work even if there is non-ascii, but you can easily test that in (i)python |
| 18:55 | <zcorpan> | >>> re.search(r'q=([^&]+)', b'q=\xE5').groups() |
| 18:55 | <zcorpan> | ('\xe5',) |
| 18:56 | <zcorpan> | hmm i guess that wasn't what i wanted to test |
| 18:56 | jgraham | -> away for a bit |
| 19:15 | <hsivonen> | annevk-cloud: a quick looks at an Armenian legacy site suggests that IE doesn't do what Gecko does with x-user-defined, either |
| 19:16 | <hsivonen> | this sort of knowing Unicode-avoidance legacy is sad |
| 19:16 | <hsivonen> | for those following along at home, I'm looking at http://archive.aravot.am/2005/aravot_arm/February/25/aravot_index.htm |
| 19:18 | <hsivonen> | in any case, we should probably re-evaluate x-user-defined |
| 19:18 | <hsivonen> | and see if making it a label of windows-1252 would lead to better interop |
| 19:22 | <zcorpan> | jgraham: urllib.urlopen(b'...\xE5').read() sends %E5 in the request |
| 19:22 | <zcorpan> | hsivonen: do webkit/ie support x-user-defined in XHR? |
| 19:23 | <hsivonen> | zcorpan: dunno. finding out would be part of re-evaluating |
| 19:24 | <SimonSapin> | hsivonen: x-user-defined is a documented trick to get binary data through XHR without arraybuffer support: https://developer.mozilla.org/en-US/docs/Web/API/XMLHttpRequest/Using_XMLHttpRequest#Handling_binary_data |
| 19:25 | zcorpan | finds https://developer.mozilla.org/en-US/docs/Web/API/XMLHttpRequest/Using_XMLHttpRequest |
| 19:25 | <zcorpan> | SimonSapin beat me to it |
| 19:53 | <zcorpan> | jgraham: writing \xe5 to a socket directly worked. it round-tripped OK |
| 20:13 | <BenoitRen> | Hi everyone. I'm still trying to mark up video game scripts, and I've hit another snag. |
| 20:13 | <BenoitRen> | Suppose I have an NPC you can talk to. I'd mark it up like this: <h2>NPC name</h2><ol><li>something</li><li>something else</li></ol> |
| 20:14 | <BenoitRen> | Where the list items are the different lines you get. |
| 20:14 | <BenoitRen> | Now, what do I do if an action occurs during or after one of those lines? |
| 20:15 | <zewt> | ... what |
| 20:15 | <BenoitRen> | For example, there's a dog NPC you have to catch. You can interact with it twice before it runs off to another town. |
| 20:15 | <zewt> | html isn't a script markup language. heh |
| 20:16 | <BenoitRen> | @zewt: I know that. But if I'm going to present it on a web page, I have to mark it up somehow. |
| 21:48 | <zcorpan> | does firefox support ping=""? |
| 21:48 | <zcorpan> | i can't see it making a request |
| 21:50 | <zcorpan> | but the IDL attribute is present :-( |
| 21:52 | <smaug____> | zcorpan: IIRC it doesn't |
| 21:53 | <smaug____> | there is still some code for the old proposal, but it is disabled by default |
| 21:53 | <smaug____> | (IMHO ping isn't too useful thing to add to the web platform) |
| 21:53 | <zcorpan> | it's sad that the .ping property is there but the feature doesn't work. feature detection fail |
| 21:54 | <zcorpan> | (also non-conforming) |
| 21:55 | <smaug____> | it sounds like a good thing. Some tracking company thinks it works, and doesn't even try to use more heavy weight solutions :) |
| 21:55 | <smaug____> | but I don't know why ping is there |
| 21:55 | <smaug____> | looking |
| 21:56 | <zcorpan> | more likely tracking companies will notice that it doesn't work and UA sniff for firefox, so when you do implement it you'll still have as bad user experience as without it :-) |
| 21:57 | zcorpan | poof |
| 21:57 | <smaug____> | anyhow, you can blame Google about this implementation |
| 21:58 | <smaug____> | http://bonsai.mozilla.org/cvsblame.cgi?file=mozilla/dom/public/idl/html/nsIDOMNSHTMLAnchorElement2.idl&rev=1.1 |
| 22:20 | <zcorpan> | smaug____: ok |
| 22:21 | <zcorpan> | smaug____: did you file a bug to remove it? |
| 22:21 | <smaug____> | nope |
| 22:22 | <smaug____> | I wonder what all will break if it is removed |
| 22:28 | <zcorpan> | might be worth checking how google mobile search detects ping support, if it's just UA sniffing |
| 22:28 | <zcorpan> | good night |