| 00:01 | <TabAtkins> | AryehGregor: I've seen the suggestion (from someone competent, iirc, maybe even a patent lawyer?) that it would be good for the world if you started a reciprocal patent pool like you describe (free to join, nobody in it can sue anybody in it, you can late-join for protection), fill it with patent trolls, and give them a bankroll to start suing people. |
| 00:02 | <TabAtkins> | It's good business for the patent trolls, and their energy has a good chance of subverting a large part of the damage that patents cause. |
| 00:03 | <zewt> | does that really matter if it's a company dedicated to litigation holding a patent? they don't care about retaliation, since they're not doing anything themselves to violate other patents |
| 00:04 | <TabAtkins> | The courts are, I think, gradually being better about that sort of thing. This would stop actual companies from suing each other, though. |
| 00:04 | <TabAtkins> | At least, that's the impression I get from absorbing Techdirt. |
| 00:07 | <Hixie> | are MIME parameter names required to be treated case-insensitively? |
| 00:07 | <Hixie> | or is that something i have to say each time? |
| 00:42 | <Sho_> | Hi folks, does anybody know what the current state of html5lib for Python 3 is? |
| 00:43 | <TabAtkins> | Sho_: You want to ping jgraham, probably. He's asleep currently, since he lives in Norway. |
| 00:44 | <Sho_> | TabAtkins: thanks, I'll try to catch him during the CE(S)T day |
| 00:45 | <gsnedders> | TabAtkins: s/Norway/Sweden/ |
| 00:45 | <gsnedders> | Sho_: It's unmaintained. |
| 00:45 | <TabAtkins> | gsnedders: Same thing. |
| 00:45 | <gsnedders> | TabAtkins: I'm here, he isn't. QED. |
| 00:46 | <gsnedders> | Sho_: It's an old 2to3 based port of the python2 version |
| 00:46 | <Sho_> | gsnedders: yeah, that's what I feared when i saw the 2009 modification dates in the folder :-) |
| 00:47 | <gsnedders> | Sho_: Should probably try and start moving towards getting it working again… |
| 00:47 | <Sho_> | gsnedders: Do you have a handle on how 2to3 was missing and how much work remains? I have a html5lib-using codebase that I'd love to get on py3k right now; depending on the size of the project I might be able to scratch my itch there |
| 00:48 | <Sho_> | s/how 2to3/how much 2to3/ |
| 00:48 | <gsnedders> | Sho_: I don't knoq. |
| 00:48 | <gsnedders> | *know |
| 00:49 | <Sho_> | alright, then I'll get a stick and poke it myself a bit |
| 00:49 | <gsnedders> | Sho_: At least now there's quite a few strings that should be unicode strings that aren't |
| 00:50 | <gsnedders> | Sho_: So you can probably get somewhere tidying stuff up in the Py2 code to make 2to3 work better |
| 00:51 | <Sho_> | gsnedders: what's the lowest version of python html5lib officially supports? if it's OK to depend on 2.6, it might be a good first step to get it to work on from __future__ import unicode_literals to increase the amount of code that transfers over unchanged |
| 00:52 | <Sho_> | but I assume you're targetting 2.4 /2.5 still |
| 00:53 | <gsnedders> | Sho_: I dunno. We tend to support the oldest anyone bothers testing. I'd rather we kept support for 2.5, at least. |
| 00:54 | <Sho_> | OK, then it's more reasonable indeed to make changes catering to making 2to3 more effective |
| 00:55 | <gsnedders> | Sho_: unicode_literals I'd rather not rely upon, but it may be nice to use while testing the Py2 copy locally to find more needing changed |
| 00:56 | <Sho_> | yep |
| 01:00 | <gsnedders> | Sho_: Anything you do change file bugs/patches on |
| 01:00 | <Sho_> | gsnedders: roger, I'll try to keep things to reviewable sizes |
| 01:01 | Sho_ | has to talk to his employer first, but it would be very neat if i could actually spend some company time on this |
| 01:35 | <Hixie> | jgraham: got a 504 this time :-) |
| 03:31 | <MikeSmith> | http://www.contextis.com/resources/blog/webgl2/ |
| 03:31 | <MikeSmith> | wow |
| 03:31 | <MikeSmith> | "we show how anyone running Firefox 4 with WebGL support is vulnerable to having malicious web pages capture screenshots of any window on their system" |
| 06:22 | <roc> | I love the spam in public-html |
| 06:23 | <roc> | "WHAT ARE YOU DOING MAN TAKING SO MUCH TIME ? YOU GUYS SHOUTED MORE AND GIVE |
| 06:23 | <roc> | OUTPUT LITTLE?? YOU CAN LEARN FROM ADOBE ORGANIZATION!" |
| 06:27 | <MikeSmith> | heh |
| 06:27 | <MikeSmith> | roc: yeah, that one is a gem indeed |
| 06:28 | <MikeSmith> | I think we should make that the /topic for this channel |
| 06:30 | <MikeSmith> | roc: btw, your latest blog posting was pretty intriguing |
| 06:31 | <MikeSmith> | I hope you'll keep posting updates |
| 06:31 | <MikeSmith> | recent discussions on the audio WG mailing list have pretty interesting also |
| 06:57 | <nessy> | couldn't agree more - on both accounts! That "bug" made me smirk, too :-) |
| 07:35 | <hsivonen> | what does this mean? http://www.tbray.org/ongoing/When/201x/2011/06/14/Native-vs-Web#c1308139244.984870 |
| 07:40 | <MikeSmith> | hsivonen: I suspect he defines "the Web" as something different than we do |
| 07:40 | <MikeSmith> | similar to the way that Jukka defines "validation" and being DTD-based |
| 07:41 | <hsivonen> | MikeSmith: Tim Bray defines Web app in a way that doesn't match the way most people use the term |
| 07:42 | <MikeSmith> | not surprised |
| 07:42 | <hsivonen> | Word can load .doc straight off an HTTP URL, right? so that makes Word a Web app per Tim Bray's definition |
| 07:43 | <MikeSmith> | heh |
| 07:43 | <MikeSmith> | well, hard to know what to say positively about that |
| 07:43 | <MikeSmith> | I guess that makes Word and HTTP UA |
| 07:44 | <MikeSmith> | which means it's just as important as browsers are |
| 07:45 | <MikeSmith> | so we should give up our misguided focus on browsers and realize that their are a myriad of equally important class of other applications to concern ourselves with |
| 07:45 | <MikeSmith> | anyway, I suspect that "entirely incompatible with the Web" probably means not in line with whatever specs they imagine as defining how the Web is supposed to work |
| 07:45 | <MikeSmith> | instead of how it actually works in practice |
| 07:46 | <MikeSmith> | I think some people really just hate the Web |
| 07:46 | <MikeSmith> | the Web as it actually exists |
| 07:46 | <MikeSmith> | because it has turned out messy and ugly |
| 07:46 | <MikeSmith> | and they wanted it all to be purty |
| 07:46 | <MikeSmith> | and neat |
| 07:46 | <MikeSmith> | and I think that's part of what they really hate about the HTML5 spec |
| 07:47 | <MikeSmith> | that is, the HTML5 spec documents the real Web |
| 07:47 | <MikeSmith> | the messy, ugly one |
| 07:47 | <Ms2ger> | We should replace it with something purty and neat, all from scratch |
| 07:47 | <MikeSmith> | we should just follow the specs |
| 07:47 | <MikeSmith> | then everything would be fine |
| 07:47 | <MikeSmith> | the good specs, I mean |
| 07:48 | <MikeSmith> | the pure "the Right Thing" ones |
| 07:48 | <Hixie> | you mean the ones that don't define error handling which is how we end up in this mess in the first place? :-) |
| 07:48 | <Ms2ger> | Yes, who needs error handling? |
| 07:49 | <Ms2ger> | Authors should just get it right all the time |
| 07:50 | <Hixie> | writing successful good specs is hard. |
| 07:51 | <Hixie> | some of the most successful things in the specs i write are full of quirks because i made mistakes while speccing them and by the time they were noticed, implementations had shipped and we were stuck. |
| 07:51 | <Hixie> | better that than a spec that's not implemented, though |
| 08:14 | <hsivonen> | oh well. I commented on tbray's blog post. Accidental 386 time. |
| 08:40 | <Dashiva> | hsivonen: Maybe one day you'll be able to afford an upgrade to 486 :) |
| 08:41 | <hsivonen> | Dashiva: to not being a ninja? |
| 08:44 | <Dashiva> | hsivonen: At least that way you'll get to sleep at night |
| 09:06 | <zcorpan> | Hixie: clearly you need some summary and scope and abbr and axis and longdesc on that table |
| 09:25 | <MikeSmith> | this party needs some annevk |
| 09:58 | <zcorpan> | MikeSmith: you're far too pragmatic! |
| 09:58 | <MikeSmith> | ? |
| 09:59 | <zcorpan> | bug 339 |
| 10:02 | <MikeSmith> | ah |
| 10:03 | <MikeSmith> | i find that code not so much fun to work with :) so I think it best to minimize how much of it to touch |
| 10:05 | <MikeSmith> | Hixie: I, um, just accidentally checked in some changes to the W3C sources for the spec |
| 10:05 | <MikeSmith> | so if you get conflicts next time you push, please just blow away what's there |
| 10:06 | <MikeSmith> | oh |
| 10:06 | <MikeSmith> | nm |
| 10:06 | <MikeSmith> | actually, it seems like I didn't do what I thought I did |
| 10:07 | <MikeSmith> | so I'm not as stupid as I thought |
| 10:07 | <MikeSmith> | yay me |
| 10:07 | <nessy> | lol |
| 10:34 | <hsivonen> | I'm shocked. I found a bug in the V.nu/Firefox parser impl. while prototyping Complex Ruby support |
| 10:34 | <hsivonen> | the bug was in the "any other end tag case" |
| 10:37 | <MikeSmith> | cool |
| 10:37 | <MikeSmith> | so Complex Ruby has finally proven useful for something |
| 10:53 | <zcorpan> | what's teh bug? |
| 10:56 | <jgraham> | Transposed e and h? |
| 10:57 | <zcorpan> | "In Section 7, the text about the close /reason/ makes it sound as if an |
| 10:57 | <zcorpan> | application might choose to show UTF-8 encoded data to an end user. That |
| 10:57 | <zcorpan> | might lead the reader to think that language tagging might be necessary. |
| 10:57 | <zcorpan> | Is it?" |
| 10:58 | <jgraham> | zcorpan: huh? |
| 10:58 | <zcorpan> | hybi |
| 10:59 | <zcorpan> | let's localize the close frame reason |
| 10:59 | <zcorpan> | maybe the reason needs some metadata as well |
| 11:00 | <jgraham> | Oh wow |
| 11:01 | <jgraham> | I didn't even consider that they might be saying that |
| 11:02 | <jgraham> | We could allow multiple /reason/s and have RDF to express equivalence relationships between them |
| 11:02 | <jgraham> | So that people could be clear that reason 2 was reason 1 translated into esperanto |
| 11:02 | <zcorpan> | RDF in JSON you mean? |
| 11:07 | <Ms2ger> | Is that the new Godwin? |
| 11:08 | <hsivonen> | zcorpan: it generated implied end tags without excluding the name of the token |
| 11:09 | <hsivonen> | zcorpan: there was also a check to see if the current node matched the token before the loop that's now in the spec |
| 11:09 | <hsivonen> | I suspect there has been a spec bug and I've failed to update the implementation when the spec got fixed |
| 11:09 | <zcorpan> | yeah i recall a spec change in that area some time ago |
| 11:10 | <zcorpan> | Philip`: plz fix the spec splitter around video |
| 11:13 | <hsivonen> | zcorpan: do you happen to know why rt and rp behavior is conditional to ruby being in scope? |
| 11:13 | hsivonen | launches a Windows VM to examine the ruby behavior of old IE |
| 11:14 | <hsivonen> | wow. |
| 11:14 | <hsivonen> | old IE treats rt as unknown unless inside ruby |
| 11:15 | <hsivonen> | that's even weirder than what I expected from old IE |
| 11:15 | <zcorpan> | hsivonen: you mean outside ruby (for old ie)? |
| 11:15 | <zcorpan> | oh |
| 11:15 | <zcorpan> | i missed unless |
| 11:16 | <zcorpan> | hsivonen: i recall having asked the same question to Hixie |
| 11:17 | <zcorpan> | i don't remember what the answer was exactly |
| 11:18 | <zcorpan> | http://krijnhoetmer.nl/irc-logs/whatwg/20100617#l-1194 |
| 11:19 | <hsivonen> | zcorpan: thanks |
| 11:19 | <hsivonen> | I guess I don't care about the scope check either way |
| 11:20 | <hsivonen> | Looks like Drupal owns the CMS that does RDFa market |
| 11:21 | <jgraham> | hsivonen: I entirely failed to understand that sentence |
| 11:27 | <hsivonen> | jgraham: Drupal owns the market that is scoped as "CMS that does RDFa" |
| 11:33 | <annevk> | oops, did not intend to open this quite yet, oh well |
| 11:33 | <Ms2ger> | Ohai |
| 11:34 | <annevk> | I wonder if Opera will crash due to the amount of email |
| 11:34 | <zcorpan> | WB |
| 11:34 | <annevk> | thanks |
| 11:35 | <Ms2ger> | ^That |
| 11:35 | <zcorpan> | shelley stopped doing whatwg weeklies |
| 11:36 | <hsivonen> | annevk: welcome back |
| 11:36 | <annevk> | zcorpan, I'll aim for Monday |
| 11:36 | <hsivonen> | hmm. fantasai isn't here |
| 11:36 | <zcorpan> | cool |
| 11:36 | <annevk> | at the moment email is not fetched beyond March 22... |
| 11:36 | <hsivonen> | does anyone have any idea why fantasai proposed putting <rp> inside <rtc>? |
| 11:37 | <MikeSmith> | annevk: stuff has gone all to hell during the last three months. please fix everything |
| 11:37 | <annevk> | hsivonen, she wrote a blog post no? |
| 11:37 | <hsivonen> | annevk: yes |
| 11:37 | hsivonen | re-reads again |
| 11:37 | <annevk> | hmm |
| 11:37 | <MikeSmith> | hsivonen: I thought fantasai was just following whatever was in the original (complex) Ruby spec |
| 11:37 | <annevk> | fetching 323 / 17091 |
| 11:38 | <MikeSmith> | you're a brave man for using Opera for e-mail |
| 11:39 | <hsivonen> | annevk: I re-read again. I still don't see why one would make rp a child of rtc |
| 11:39 | <hsivonen> | MikeSmith: I don't see this case in the original spec |
| 11:39 | <MikeSmith> | oh |
| 11:40 | <hsivonen> | sigh. I guess I will have to ask fantasai before requesting specific spec edits |
| 11:43 | <annevk> | I kind of thought we decided not to address all ruby scenarios for now, but oh well |
| 11:44 | <annevk> | I guess no it is semi-implemented... |
| 11:45 | <hsivonen> | annevk: the point here is to avoid painting ourselves in the corner with an over-aggressive parsing algorithm |
| 12:39 | <zcorpan> | heycam|away: any chance to get a decision on http://www.w3.org/Bugs/Public/show_bug.cgi?id=12798 soon? |
| 12:45 | <zcorpan> | Hixie: http://www.whatwg.org/specs/ doesn't list all whatwg-hosted specs |
| 12:46 | <zcorpan> | or maybe it does |
| 12:51 | <annevk> | we're not going to do null -> "" after all? |
| 12:51 | <annevk> | pfff |
| 12:54 | <zcorpan> | seems so |
| 12:54 | <annevk> | btw, compared to the 11'' Air the MacBook Pro 13'' is huge |
| 12:55 | <annevk> | and heavy |
| 12:57 | <hasather> | annevk: hey, welcome back. When are you back in Oslo? |
| 13:04 | <asmodai> | saw this? |
| 13:04 | <asmodai> | http://www.unifont.org/tt/ |
| 13:08 | <MikeSmith> | asmodai: the word "elegant" does not come to mind when looking at that solution |
| 13:09 | <annevk> | hasather, no idea really, apparently not in time for the party |
| 13:12 | <asmodai> | MikeSmith: I wonder how much work is still needed to get it all properly working natively. |
| 13:14 | <MikeSmith> | I guess what's needed for proper rendering of vertical text is for browsers to actually support proper rendering of vertical text |
| 13:14 | <MikeSmith> | without using transform and rotate |
| 13:16 | <MikeSmith> | like this: |
| 13:16 | <MikeSmith> | http://people.w3.org/mike/demo/melos/ |
| 13:18 | <MikeSmith> | using writing-mode |
| 13:23 | <asmodai> | Pressing x doesn't seem to work for me though. |
| 13:23 | <MikeSmith> | in only works in browsers that support writting-mode |
| 13:24 | <MikeSmith> | which I think means only Safari and Chrome at this point |
| 13:25 | asmodai | fires up chrome |
| 13:26 | <annevk> | I saw something go by about the CSS WG considering once again Unicode normalization |
| 13:26 | <annevk> | did that get resolved the right way? |
| 13:26 | <asmodai> | ok, it rotates 90 degrees. |
| 13:26 | <annevk> | guess I'll find out once I get through my email backlog |
| 13:27 | <asmodai> | But then again the kanji, hiragana, katakana als rotate, but hey, you know enough about Japanese to know what the end result should become. |
| 13:27 | <MikeSmith> | annevk: yeah, Richard told me a few weeks ago they were having some discussion, but I don't now what the resolution was |
| 13:27 | <MikeSmith> | if there was any |
| 13:29 | <MikeSmith> | asmodai: I think everything gets displayed the way it should when that is displayed in vertical mode -- with the exception of the roman characters |
| 13:29 | <MikeSmith> | that is, the URL at the end |
| 13:29 | <MikeSmith> | or maybe that's actually per spec |
| 13:29 | <MikeSmith> | I don't really know what's supposed to happen with roman characters by default when in vertical mode |
| 13:30 | <asmodai> | on Chrome at least all the kanji are rotated as well. If it's supposed to act like Japanese writing I'd expect the kanji, hiragana, katakana to have their orientation intact instead of also being rotated 90 degrees. |
| 13:31 | <asmodai> | latin text tends to get rotated 90 degrees |
| 13:31 | <asmodai> | so bottom faces left hand side |
| 13:31 | <MikeSmith> | the kanji and kana display as expected in my Chrome |
| 13:31 | <asmodai> | MikeSmith: which version? |
| 13:31 | <MikeSmith> | which are current dev channel and canaray |
| 13:31 | <asmodai> | that's odd |
| 13:32 | <asmodai> | I'm following Chrome dev as well (14) |
| 13:32 | <MikeSmith> | 14 974 is what I have |
| 13:32 | <asmodai> | 794 here |
| 13:32 | <MikeSmith> | wacky |
| 13:33 | <MikeSmith> | well, the kanji and kana should definitely not rotate |
| 13:34 | <annevk> | hmm |
| 13:34 | <annevk> | last email on public-web-notification was on May 31 |
| 13:34 | <asmodai> | MikeSmith: http://www.in-nomine.org/~asmodai/whatiget.png |
| 13:34 | <annevk> | a call for exclusions... |
| 13:35 | <MikeSmith> | asmodai: hmm, yeah, it ain't supposed to look like that :) |
| 13:36 | <asmodai> | MikeSmith: Odd that you get what it should be on the same Chrome version :| |
| 13:36 | <asmodai> | MikeSmith: any options that might be enabled on your side? |
| 13:36 | <MikeSmith> | none that I recall |
| 13:37 | <MikeSmith> | asmodai: could be a platform difference |
| 13:38 | <MikeSmith> | I am running on OSX |
| 13:38 | <MikeSmith> | maybe it relies on some mac-ish thingy |
| 13:38 | <asmodai> | same here |
| 13:38 | <asmodai> | ah |
| 13:38 | <asmodai> | Windows 7 here |
| 13:38 | <MikeSmith> | so that's probably it |
| 13:38 | <MikeSmith> | dunno |
| 13:39 | <MikeSmith> | I need to get a WIndows VM set up |
| 13:39 | <asmodai> | checking safari |
| 13:40 | <hsivonen> | Today is a good day to follow both @diveintomark and @JeniT on Twitter |
| 13:41 | <asmodai> | MikeSmith: just checked |
| 13:41 | <asmodai> | MikeSmith: colleague just checked on his Mac with Chrome 14 |
| 13:41 | <asmodai> | MikeSmith: works for him as well, it's a platform issue |
| 13:42 | <MikeSmith> | ok |
| 13:42 | <asmodai> | Anyway, that unifont thing was raised on the unicode list. |
| 13:42 | <erlehmann> | twitter, wasn't that the service breaking the web with shortlinks, killing feeds and using hashbangs instead of history API? |
| 13:42 | erlehmann | snickers |
| 13:44 | <asmodai> | Speaking of which, Gawker seems to have cleaned up their act. |
| 13:44 | jgraham | is developing a great deal of respect for Jeni T even though we have very different taste in technologies |
| 13:45 | <jgraham> | And despite the fact that she is trying to have a sensible conversation over twitter |
| 13:45 | <jgraham> | Which is pretty silly… |
| 13:48 | <gsnedders> | jgraham: There was a reason why I was sad when she decided not to run for election to the TAG… and then happy when she got appointed by timbl anyway. :) |
| 13:49 | <jgraham> | Well on the whole I would prefer the TAG to be populated by reasonable people with good taste in technologies ;) |
| 13:50 | <gsnedders> | I dunno — it's probably too heavily occupied by semweb people (which is a not unreasonable direction to take, and used quite heavily for interchange) — but getting anyone else willing to be on it is challenging… |
| 13:52 | <annevk> | Why do we need the TAG again? |
| 13:53 | <gsnedders> | annevk: Not having multiple WGs trying to do the same thing is in general a good idea. And people would probably complain if there wasn't a group whose job it was to advise the Director. |
| 13:54 | <hsivonen> | gsnedders: part of the problem with not having multiple WGs is that it's too easy for the first WG on a given topic to do the wrong thing |
| 13:54 | <jgraham> | Not having multiple working groups is part of the problem not the solution |
| 13:55 | <gsnedders> | How does have multiple distinct groups writing competing specs help the market? |
| 13:56 | <jgraham> | It gives the market a chance to choose the spec that fits its needs best |
| 13:56 | <jgraham> | Rather than assuming the the needs of the people writing the spec are well matched to those of the market |
| 13:58 | <gsnedders> | jgraham: So fragmentation like HD DVD/BluRay is good? |
| 13:59 | <jgraham> | Sure |
| 13:59 | <jgraham> | You notice there is no fragmentation now |
| 14:00 | <hsivonen> | there is if you look beyond plastic discs, though |
| 14:01 | <gsnedders> | I'd argue the fragmentation hurt the market then — it slowed the uptake of HD disks in general as far as I can tell |
| 14:02 | <hsivonen> | gsnedders: assuming that dealing with discs is not hurting |
| 14:02 | <hsivonen> | gsnedders: maybe scoping the problem as discs is as wrong as scoping the problem as how to serialize RDF |
| 14:02 | <Lachy> | gsnedders, the market is trying to move beyond physical media, which also partly explains the slower uptake |
| 14:03 | <gsnedders> | hsivonen: I think the problem with RDFa/Microdata is people trying to say that RDFa already exists for HTML. |
| 14:03 | <jgraham> | gsnedders: I think the slow uptake could be because people just didn't want what either format offered |
| 14:03 | <gsnedders> | jgraham: It seems to be increasing now, though |
| 14:03 | <Lachy> | and that market is in turn hurt by lower bit-rates, incompatible and annoying DRM systems, ISP download caps, etc |
| 14:03 | <jgraham> | Yes, and so is the affordability of HD TVs |
| 14:03 | <hsivonen> | gsnedders: the Process bogosity and the manufacture of legacy are a problem--not *the* problem |
| 14:04 | <hsivonen> | Lachy: shoddy software, territorial restrictions, etc., etc. |
| 14:05 | <hsivonen> | at home, I have a separate boot partition for online video rental to make sure the software doesn't come in contact with the daily use system |
| 14:06 | <hsivonen> | and the reason to bother with something that makes one feel the need to wall off the video rental software from other software is that the service that ties into better software isn't available in my region |
| 14:07 | <Lachy> | which service do you use? |
| 14:08 | <hsivonen> | Lachy: Voddler |
| 14:10 | <Lachy> | oh, I haven't heard of that one. What software does it use? Flash or Silverlight? |
| 14:10 | <Lachy> | or other? |
| 14:10 | <hsivonen> | Lachy: Flash Player for playback. |
| 14:11 | <hsivonen> | Lachy: custom networking |
| 14:11 | <hsivonen> | they used to use Adobe AIR |
| 14:11 | <hsivonen> | AIR, of course, is so well Integrated that it performs worse that in-browser Flash Player |
| 14:13 | <hsivonen> | would be nice if they used browser-native WebM and HTTP |
| 14:13 | <hsivonen> | but they have the DRM-wanting content provider overlords |
| 14:13 | <hsivonen> | also, they seem to think using HTTP would be too expensive |
| 14:14 | <Lachy> | reading their help, they said their using some peering technology that they developed. I guess that's something like streaming torrents |
| 14:14 | <jgraham> | """you can't imagine all the |
| 14:14 | <jgraham> | hacks browser manufacturers do to repair really bad code""" |
| 14:15 | <jgraham> | Umm, well actaully we can |
| 14:15 | <jgraham> | See also the HTML5 spec |
| 14:15 | <gsnedders> | Source? |
| 14:15 | <jgraham> | http://www.w3.org/mid/OF31B33356.F668E1E0-ON862578B2.00485AD0-862578B2.00489BBD⊙uic |
| 14:21 | richt | until someone meets my demands at dontmakemesteal.com I'll be on cuevana.tv amigos...HTML video would be nice once we've got a full screen playback API sorted. |
| 14:31 | <erlehmann> | gsnedders, RDFa exists. creative commons, for example, is using it. |
| 14:31 | <erlehmann> | but really i have no idea of the extent of use of RDFa vs microdata markup. |
| 14:32 | <jgraham> | In general people publishing stuff is not interesting if no one consumes it |
| 14:33 | <erlehmann> | i see what you did there. schema.org comes to mind. |
| 14:34 | <hsivonen> | erlehmann: creative commons is a particularly sad case considering that RDF is such an overkill for license annotations |
| 14:34 | erlehmann | knows what he did last summer. |
| 14:36 | <Lachy> | does creative commons still enocurage people to publish RDF inside <!-- comments -->? |
| 14:36 | <gsnedders> | Lachy: RDF/XML? |
| 14:36 | <erlehmann> | go ask in #cc |
| 14:36 | <erlehmann> | i don't think so. that'd be awkward. |
| 14:45 | <Lachy> | gsnedders, yeah, they used to do something like that. http://www.xml.com/pub/a/2003/01/15/creative.html |
| 14:45 | <gsnedders> | Lachy: I know WP did. Didn't know about what CC suggested. |
| 14:46 | <gsnedders> | Lachy: But, well, something like RDFa within a comment would make even less sense |
| 14:47 | <hsivonen> | gsnedders: the WHATWG licensing Microdata vocab is remarkably sensible compared to ccREL |
| 14:48 | <hsivonen> | ooh. Microsoft now explicitly considers WebGL harmful: http://blogs.technet.com/b/srd/archive/2011/06/16/webgl-considered-harmful.aspx |
| 14:51 | <AryehGregor> | TabAtkins_, that patent troll pool idea was actually mine originally. Florian Mueller later picked it up and posted about it on his blog: http://fosspatents.blogspot.com/2010/05/dpl-and-fair-troll-business-model-make.html |
| 14:52 | <AryehGregor> | But I don't think he followed up on it, or at least I don't know about it. |
| 14:53 | <AryehGregor> | zewt, if all the big companies were unable to enforce their patents against competitors, it would only be patent trolls supporting patents, so the big companies would lobby successfully to reform patent law in some fashion. |
| 14:53 | <AryehGregor> | (possibly just by making such patent pools illegal, though, if they aren't already) |
| 14:54 | <AryehGregor> | hsivonen, well . . . to be fair, they have a point. It's pretty scary. |
| 14:55 | <jgraham> | AryehGregor: More than the silverlight 3D thing? |
| 14:55 | <hsivonen> | AryehGregor: I'm not saying they don't have a point. |
| 14:55 | <AryehGregor> | jgraham, it's entirely possible that the people who reviewed WebGL and found it insecure didn't even know about Silverlight's 3D features, and can't do anything about them because it's a different department. |
| 14:56 | <AryehGregor> | Like how Microsoft officially opposed @font-face without obfuscation when Silverlight supported something similar already. |
| 14:56 | <AryehGregor> | (then their objection suddenly became that only *declarative* unobfuscated font use was a problem) |
| 14:56 | <hsivonen> | what's the level of abstraction that Silverlight's 3D stuff provides? |
| 14:56 | <hsivonen> | what about Molehill? |
| 14:57 | <jgraham> | Entirely possible. But it's still pretty hypocritical to publically oppose a technology on the one hand and push an identical (?) technology on the other |
| 14:58 | <AryehGregor> | You can't fairly call an 89,000-employee critical because some of its employees contradict other employees. |
| 14:59 | <AryehGregor> | Yay: http://www.imperialviolet.org/2011/06/16/dnssecchrome.html |
| 15:00 | <jgraham> | Well I can because that's what hypocritical means. It might not be possible, or even desirable, for a large company to avoid hypocritsy however |
| 15:00 | <AryehGregor> | No, hypocrisy is when someone pretends to be something they aren't, or says one thing and does another. An organization can't be hypocritical, only people can be hypocritical. |
| 15:01 | <AryehGregor> | Even if we sometimes treat organizations like people. |
| 15:02 | <jgraham> | AryehGregor: I quite disagree |
| 15:03 | <jgraham> | I suppose the other reading is that "In its current form, WebGL is not a technology Microsoft can endorse from a security perspective" means that Microsoft doesn't consider this level of security concerns a sufficient reason not to release a product |
| 15:03 | <jgraham> | (or that the silverlight thing actually has materially different security properties) |
| 15:04 | <AryehGregor> | No, it probably just means that totally different people did the security reviews. |
| 15:04 | <Philip`> | It seems reasonable to assume Microsoft has divisions that apply different weights to different criteria - their Security Response Center is clearly going to focus on security and recommend against insecure things, while their Silverlight people and IE people etc will focus more on enhancing the customer experience through hardware acceleration |
| 15:04 | <AryehGregor> | From different groups that maybe barely talk to each other. |
| 15:04 | <jgraham> | AryehGregor: But they claim to speak for Microsoft as a collective entity |
| 15:04 | <AryehGregor> | Well, yeah, that's a common fiction in large organizations. |
| 15:04 | <jgraham> | Not Microsoft's security team or anything else |
| 15:05 | <AryehGregor> | It's quite possible that the Silverlight team needs to get adoption at all costs, because if they fail to get adoption for long enough they might be axed or at least lose money to other groups. |
| 15:05 | <AryehGregor> | So maybe they aren't so interested in others concerns. |
| 15:05 | <AryehGregor> | IE is going to get lots of resources regardless, is much more widely used, is under more public scrutiny, and has often been criticized for poor security. |
| 15:05 | <AryehGregor> | Totally different situation. They might be a lot less willing to take security or other risks. |
| 15:08 | <jgraham> | Indeed. But it's still hypocritsy if "Microsoft" announce that it is unacceptable to ship a technology for reason X whilst shipping a similar technology to which X also applies |
| 15:10 | <AryehGregor> | No, it's just people claiming to speak for Microsoft when they really shouldn't be. |
| 15:10 | <hsivonen> | "For security and compatibility reasons, some drivers will be blocked by default in the browser." |
| 15:11 | <hsivonen> | http://msdn.microsoft.com/en-us/library/gg197424%28v=XNAGameStudio.35%29.aspx |
| 15:11 | <AryehGregor> | It's not like the guy who posted that got approval from Steve Ballmer to speak in the name of Microsoft. |
| 15:47 | <krijnh> | Nothing to see here, move along |
| 15:50 | <annevk> | oh dictionaries in Web IDL |
| 15:52 | <krijn> | annevk: power supply died, nothing aan de hand :) Maar zal meteen weer wat backups overzetten |
| 15:53 | <annevk> | #panicmode |
| 15:54 | <krijn> | Also, we're getting fiber here, so massive speed improvement coming up! |
| 16:00 | <zcorpan> | rwaldron: sorry gotta go now. back on monday. feel free to email me simonp⊙oc |
| 16:00 | <rwaldron> | ok, thank |
| 16:00 | <rwaldron> | thanks* |
| 16:02 | <annevk> | so no HTML/XML work done since I left |
| 16:02 | <annevk> | not sure what is up with URL |
| 16:02 | <annevk> | no discussion on Web Notifications |
| 16:38 | <krijnh> | Why, what's up? |
| 16:38 | <krijn> | Shut up you |
| 16:38 | <jcranmer> | sigh |
| 16:38 | <krijn> | You're an unreliable piece of junk! |
| 16:38 | <krijnh> | :( |
| 16:38 | <jcranmer> | I wish people would implement useful things like <input type="date"> |
| 16:48 | <jgraham> | jcranmer: Now you made us feel unloved :( |
| 16:49 | <annevk> | jcranmer, Opera did, I hear iOS5 does too... |
| 16:49 | <Philip`> | jgraham: I have nearly constant logs since January 2009, conveniently split into files no larger than 55MB each |
| 16:50 | <jcranmer> | annevk: that's great for the ~2% of the people who use Opera, not so great for the 98% who don't |
| 16:50 | <annevk> | jcranmer, fix it! |
| 16:51 | <krijn> | (For those interested: http://krijnhoetmer.nl/irc-logs/ is a static file now, not live. I'm backing up everything right now, cause I have the feeling the broken power supply kind of damaged my hard drive) |
| 16:51 | <jcranmer> | unfortunately, I have no intention of touching content code |
| 16:52 | <krijn> | (Very cool btw, each time somebody requests a file from my server, I hear a high pitch sound) |
| 16:54 | <Philip`> | krijn: If you tell us the pitch that each different file makes, we could try to set up a hard disk orchestra on your machine |
| 16:54 | <annevk> | krijn, if you need hardware / money for hardware let me know |
| 16:54 | <krijn> | Nah, I have some hardware left here, that's not it (brother is buying a new power supply right now) |
| 16:55 | <krijn> | Just don't think I'm able to do the same crappy configuration on a new machine :p |
| 16:56 | <annevk> | sounds like a loss |
| 16:57 | <krijn> | Glad my mom is kind of old and doesn't hear the sound, it's pretty annoying.. |
| 16:57 | <annevk> | I wonder why we have crossorigin=""; CORS was designed so no markup changes would be needed |
| 16:58 | annevk | creates todo |
| 16:59 | <AryehGregor> | What happened, a disk failure? |
| 16:59 | <krijn> | Now, power supply overheated I think, ended with a big bang. Think it damaged something else on its way out |
| 16:59 | <krijn> | *no |
| 17:00 | <AryehGregor> | annevk, the point seems to be to allow requests even to untrusted domains in some cases, without server-side opt-in, but without submitting the user's credentials. |
| 17:00 | <AryehGregor> | Which can then be treated more leniently than other requests for stuff like canvas, I guess? |
| 17:00 | <AryehGregor> | Hmm, maybe I'm wrong. |
| 17:00 | AryehGregor | hasn't paid attention to it |
| 17:09 | <annevk> | credential-less requests still need opt-in |
| 17:10 | <annevk> | otherwise intranets guarded by IP-address protection are vulnerable |
| 17:14 | <krijn> | 7000 URIs, adactio will kill me if I screw this up :] |
| 17:15 | <annevk> | btw, does Web IDL dictionary work for arbitrary objects? |
| 17:16 | <annevk> | or will initEvent() need to be overloaded with a version for each potential dictionary? |
| 17:17 | <krijn> | (Not to mention those evil cabal assassins) |
| 17:18 | <AryehGregor> | Could we solve the intranet problem by just saying that nothing on a public IP address can load anything from a private IP address? |
| 17:19 | <AryehGregor> | Maybe some businesses put things on firewall-protected public IP addresses, but maybe we could just ignore them. |
| 17:19 | <AryehGregor> | That would make a lot of things simpler. |
| 17:19 | <AryehGregor> | Probably we can't, though. |
| 17:19 | <AryehGregor> | Might be a good idea, but maybe not good enough. |
| 17:19 | AryehGregor | can dream |
| 17:20 | <Philip`> | I don't think there's any guarantee that intranets are on private IP addresses |
| 17:20 | <AryehGregor> | No, but we could always decide not to care about the ones that aren't. |
| 17:21 | <Philip`> | We could always decide not to care about the ones that are, too |
| 17:21 | <AryehGregor> | In principle, yes. |
| 17:21 | <annevk> | that ship sailed |
| 17:22 | <AryehGregor> | It could still be changed in principle, no? |
| 17:22 | <AryehGregor> | (not that I'm optimistic) |
| 17:22 | <Philip`> | Based on zero evidence, I'd imagine that non-private-IP intranets are not significantly rarer and less worthy of care than private-IP intranets |
| 17:23 | <AryehGregor> | Most organizations have a limited number of IP addresses, and private-IP networks don't even have to be specially firewalled, so using private IP ranges is very common. |
| 17:23 | <AryehGregor> | Although it's true that organizations with lots and lots of public IP addresses (like colleges) tend to just use those for everything. |
| 17:24 | <annevk> | last I checked it was not going to happen |
| 17:25 | <annevk> | in theory everything can be changed, in practice I'm past some of those battles :) |
| 17:31 | <Philip`> | AryehGregor: Private IPs sound like a pain if you ever merge multiple intranets (companies buying each other) since the addresses may conflict, and presumably make it harder to punch holes in the firewall, so I'd expect there are reasons to use public addresses |
| 17:32 | <AryehGregor> | Presumably. |
| 17:32 | <Philip`> | (In the glorious age of IPv6 the problem of limited availability will go away) |
| 17:43 | <TabAtkins> | annevk: On the unicode normalization front, we're trying to punt, given that it's not a Namespaces issue, but i18n hasn't responded to anything lately. Chairs have scheduled a meeting with them sometime soon. |
| 17:45 | <annevk> | o_O |
| 17:45 | <annevk> | there should be no Unicode normalization |
| 17:45 | <Ms2ger> | annevk, WebIDL dictionaries can have optional properties, btw |
| 17:45 | <TabAtkins> | I don't have an opinion, but if there's any, it should happen as early as possible. And it is absolutely not an issue that the Namespaces spec needs to care about. |
| 17:46 | <AryehGregor> | It can't happen too early, because then anything that round-trips through the browser will get normalized, which is a problem if you're editing something and suddenly every line is different even though you didn't change anything. |
| 17:46 | <AryehGregor> | Generally browsers just ignore Unicode normalization and rely on the site to get it right. |
| 17:47 | <AryehGregor> | (which, e.g., MediaWiki largely does, but lots of software probably doesn't) |
| 17:47 | <TabAtkins> | Indeed. I'm just saying that *if* it happens, it's better to happen early rather than late. |
| 17:47 | <TabAtkins> | But generally, yeah, we should depend on the HTML and CSS authors to be using the same normalization. |
| 17:47 | <AryehGregor> | If "early" means during parsing or something, it can't, as I said. |
| 17:48 | <AryehGregor> | Anyway, I think this is a low-priority issue, because it only affects foreign people. If they want to use unreasonable languages, they can take the extra effort to normalize themselves. |
| 17:48 | <TabAtkins> | That's fine. My statement was a conditional. |
| 17:48 | <AryehGregor> | English speakers are largely unaffected. |
| 17:48 | <TabAtkins> | English speakers do sometimes use zalgo, though. |
| 17:49 | <AryehGregor> | Normally not for things where normalization matters much, though. |
| 17:49 | <AryehGregor> | I guess if you used a Zalgo username on a website, inconsistent normalization could cause username equality checks to fail. |
| 17:50 | <Ms2ger> | And if you do that, you deserved it |
| 17:50 | <AryehGregor> | Actually, that's a reason why browsers absolutely cannot do normalization on things like form submission: if the user's name is stored denormalized or in the wrong normalization form in the site, and the browser normalizes on submission, login would fail. |
| 17:50 | <annevk> | Ms2ger, yeah, but that is not enough |
| 17:51 | <AryehGregor> | Of course, the status quo is that login would fail if the user is logging in from multiple different platforms that do different normalization when the user types. |
| 17:51 | <AryehGregor> | Does OS X like NFD for some insane reason? |
| 17:51 | <AryehGregor> | Doesn't OS X like NFD for some insane reason? |
| 17:51 | <annevk> | Ms2ger, unless you can have ... for optional properties and let their type all be "any" |
| 17:51 | <TabAtkins> | AryehGregor: Presumably the username originally comes from a form, too, which hopefully normalizes the same way. |
| 17:51 | <Ms2ger> | Probably not |
| 17:51 | <annevk> | Ms2ger, I guess overloading is not too bad, each new event will just have to define a dictionary |
| 17:51 | <TabAtkins> | But anyway, we were talking about CSS, and yeah, we probably shouldn't do any normalization. |
| 17:52 | <AryehGregor> | TabAtkins, except right now browsers don't normalize, so if some started normalizing . . . |
| 17:52 | <TabAtkins> | Indeed. Legacy constraints, as always. |
| 17:52 | <annevk> | also, nobody uses CSS namespaces |
| 17:53 | <TabAtkins> | That too. |
| 17:53 | <AryehGregor> | They might actually be useful now with embedded SVG and MathML. |
| 17:53 | <AryehGregor> | They're one way you can feature-detect for MathML support. |
| 17:53 | <AryehGregor> | Use CSS to hide the MathML element and display an image if the MathML element has the HTML namespace. |
| 17:54 | <AryehGregor> | Not needed for SVG, though. |
| 17:54 | <erlehmann> | cool story, bro. |
| 17:54 | <erlehmann> | i did html::before once. is there a special hell for me, full of unstyled elements? |
| 17:54 | <AryehGregor> | (since it doesn't use textContent at all, so it will be invisible anyway except for fallback you can put in whatever element is suitable for that, I forget) |
| 18:02 | <MikeSmith> | hober: well stated |
| 18:06 | <annevk> | whoa, Chrome still has not disabled its crippled data:text/html,<input type=datetime> implementation? |
| 18:06 | <annevk> | madness |
| 18:06 | <krijnh> | Time for a new power supply, brb o/ |
| 18:28 | <annevk> | smaug____, about CaretSelection.range |
| 18:29 | <annevk> | it was known to be redundant with the other attributes at the time we decided to add it |
| 18:29 | <smaug____> | annevk: I don't know who decided :) |
| 18:29 | <smaug____> | IIRC, I tried to find why it was added, but couldn't |
| 18:35 | <annevk> | smaug____, http://lists.w3.org/Archives/Public/public-webapps/2009OctDec/0150.html |
| 18:36 | <smaug____> | "It could have a convenience method for converting a Range too, if that's really needed" |
| 18:36 | <smaug____> | it isn't really needed |
| 18:40 | <krijn> | Weird sound is gone \o/ |
| 18:40 | <krijn> | This might actually work |
| 18:42 | <krijn> | One change: flagging lines is only possible on the day itself, once the day is done the logfile including flagged lines is written to disk. Easier to backup/export/transfer that way.. |
| 18:43 | <krijn> | (Advantage: logs are super fast now) |
| 18:44 | <Ms2ger> | Boo ;) |
| 18:51 | <Hixie> | yay, annevk's back! wooo |
| 18:51 | <krijn> | Ow come on, what did he do for the web! |
| 18:51 | <Hixie> | jgraham: there's a ton of feature-level versioning in the spec source, just search for v[0-9] |
| 18:53 | <Hixie> | zcorpan: it lists all the actively developed specs that i have admitted are developed on whatwg :-) |
| 18:53 | <Hixie> | zcorpan: (it doesn't list xbl, which is also accessible on whatwg.org, but that's just because i needed somewhere to stick it while i developped it) |
| 18:54 | <Ms2ger> | Hixie, fyi, https://bugzilla.mozilla.org/show_bug.cgi?id=664179 |
| 18:55 | <Hixie> | looking |
| 18:55 | <Hixie> | oh yeah, i saw that |
| 18:55 | <Hixie> | it's only my list of things to do today so he has a spec to work on |
| 18:55 | <Ms2ger> | Good :) |
| 18:55 | <Hixie> | hopefully should be straight-forward |
| 19:00 | <Hixie> | yay, the w3c considers whatwg input to be internal input as far as the htmlwg goes! that's good to see. http://www.w3.org/mid/4DFB7645.4030301⊙si |
| 19:01 | <Ms2ger> | "Any comments not received by August 3 might not be considered before early 2012" |
| 19:01 | <Ms2ger> | Isn't that expected? :) |
| 19:02 | <AryehGregor> | Hixie, where does it say that? |
| 19:03 | <Hixie> | Ms2ger: dude i'm only about 3 months behind on non-RFE feedback at the moment! |
| 19:03 | <Hixie> | AryehGregor: "Group is still working through open issues of its own, outside of additional comments." |
| 19:03 | AryehGregor | doesn't know what that sentence means |
| 19:03 | <Hixie> | vast majority of work on the spec at the moment is whatwg feedback |
| 19:06 | Ms2ger | likes exaggerating... a little ;) |
| 19:06 | <mpilgrim> | heycam|away: webkit is moving to raise-on-missing-arguments in our IDL system ( see https://bugs.webkit.org/show_bug.cgi?id=62750 ), and we have plans to further align with WebIDL |
| 19:07 | <Hixie> | for RFE feedback the oldest e-mail still in the system is from 2007, so if you want to exaggerate there's plenty more room than that :-P |
| 19:08 | <Ms2ger> | I don't care about rfes |
| 19:08 | <Hixie> | fair enough |
| 19:08 | <Ms2ger> | The spec is bad enough without considering them ;) |
| 19:08 | <Hixie> | hey! |
| 19:09 | <Ms2ger> | Sure, it's the best we ever had, but you can do better ;) |
| 19:09 | <AryehGregor> | What's a case where the browser submits magical extra form fields that don't have elements associated with them? |
| 19:09 | <AryehGregor> | There are some already, right? |
| 19:09 | <Hixie> | none without elements |
| 19:09 | <Hixie> | but there are magical fields |
| 19:09 | <Hixie> | _charset_ |
| 19:09 | <Hixie> | image.x and .y |
| 19:10 | <Hixie> | can't think of any others off the top of my head |
| 19:10 | <TabAtkins> | dir |
| 19:10 | <AryehGregor> | Yeah, you added dir recently, right? |
| 19:10 | <Hixie> | oh right, some dir thing |
| 19:10 | <Ms2ger> | isindex? |
| 19:10 | <Hixie> | isindex is actually not a field :-P |
| 19:11 | <Hixie> | that's how it's magical :-) |
| 19:11 | <Ms2ger> | It's a mess, that's what it is :) |
| 19:11 | <Hixie> | no argument there |
| 19:16 | <annevk> | ah, he left |
| 19:16 | <annevk> | yay for http://diveintomark.org/archives/2011/06/17/come-on-gruber-youre-better-than-this |
| 19:17 | <annevk> | haha, the URL |
| 19:17 | <othermaciej_> | troll vs troll! |
| 19:18 | <othermaciej_> | also: welcome back annevk! |
| 19:29 | <annevk> | oh |
| 19:29 | <annevk> | so Mozilla has Nightly too... |
| 19:29 | <annevk> | what is Aurora then what I downloaded the other day? |
| 19:30 | <Ms2ger> | Between Nightly and Beta |
| 19:34 | <AryehGregor> | ARGH. |
| 19:34 | <AryehGregor> | A third post bounced, IN THE SAME THREAD. |
| 19:34 | <Ms2ger> | Time to stop posting :) |
| 19:44 | <erlehmann> | if image macros were allowed, there are quite some for STOP POSTING. |
| 19:46 | <Hixie> | annevk: btw, see <img crossorigin=""> and <video crossorigin=""> |
| 19:49 | <erlehmann> | seems the web is destined to duplicate all manners of metadata at all levels. |
| 19:49 | <erlehmann> | AryehGregor, thanks for your post on hashing. i was going to write one myself, but now I do not have to. :) |
| 19:54 | <Hixie> | AryehGregor: dude, just subscribe the other address and disable delivery |
| 19:54 | <Hixie> | AryehGregor: though at that point, why bother with a special address |
| 19:55 | <AryehGregor> | Hixie, so that if spambots get the address, I can change it and start deleting all mail to the old unused address. |
| 19:55 | AryehGregor | didn't realize you could subscribe without delivery |
| 19:56 | <AryehGregor> | Wow, there are loads of options here. |
| 20:07 | <erlehmann> | AryehGregor, i use the plus sign for additional addresses. is there another way? |
| 20:07 | <AryehGregor> | erlehmann, you can just make another address and have it forward. |
| 20:07 | <AryehGregor> | I have aryeh.name in Google Apps, solely for e-mail forwarding. I can make up to like 300 total e-mail aliases before I actually have to pay anything for it. |
| 20:08 | <AryehGregor> | But for my Gmail addresses I use pluses, yeah. |
| 20:08 | <erlehmann> | mailers gonna mail. |
| 20:16 | <Hixie> | AryehGregor: but if you're posting with your other address, surely if they get the first address they'll get teh second too? |
| 20:21 | <AryehGregor> | Hixie, well, as I've discovered, if I post to the list and it bounces, the people I send the same mail individually to reply on-list, so my sekrit address is in the CC list anyway. |
| 20:21 | <Hixie> | indeed |
| 20:21 | <Hixie> | i'd just unsubscribe the other address and use your mail one |
| 20:22 | <Hixie> | once your address is out there it's a lost cause anyway -- and having many addresses just means more spam, in my experience |
| 20:22 | <Hixie> | annevk: you around? |
| 20:22 | <AryehGregor> | Gmail tends to think some mail to the lists is spam, though, particularly from people with @google.com addresses. Currently I have a filter to not send anything to spam if it's going to my list address. |
| 20:23 | <Hixie> | annevk: CORS and EventSource -- sending cookies is pretty key, since a lot of streams are user-specific. Should I just always send them? |
| 20:23 | <AryehGregor> | Doesn't really matter much, it's just extra effort to change it at this point. |
| 20:23 | <Hixie> | AryehGregor: yeah, i often miss mail from googlers |
| 20:23 | <Hixie> | gotta love that google is making it harder for google to help google |
| 20:23 | <Hixie> | (and yet people still think i'm some sort of google puppet, lol) |
| 20:24 | <AryehGregor> | I'm not clear why. google.com has SPF, but only with softfail. What would be flagging them as spam? |
| 20:24 | <Hixie> | haven't studied it |
| 20:24 | <Hixie> | probably gmail being proactive though |
| 20:24 | <Hixie> | unless you don't use gmail? |
| 20:24 | <AryehGregor> | I have a great idea: once you're done with HTML, how about you fix SMTP? |
| 20:24 | <Ms2ger> | Hixie, of course the only reason the W3C allowed layout tables is because you're a Google puppet :) |
| 20:25 | <Hixie> | you think smtp will outlive html? there's a horrifying thought... |
| 20:25 | <TabAtkins> | I had to set a specific rule that @google.com never goes to spam. |
| 20:25 | <Hixie> | Ms2ger: yeah i love that nobody seems to complain about the obvious ibm conflict of interest on that one |
| 20:25 | <Hixie> | TabAtkins: how? |
| 20:25 | <erlehmann> | TabAtkins, that seems like fail. |
| 20:26 | <Ms2ger> | His name isn't in the spec, and nobody understand the insane HTMLWG process |
| 20:26 | <TabAtkins> | Hixie: Uh, a filter. |
| 20:26 | <TabAtkins> | erlehmann: It doesn't hurt that badly. |
| 20:26 | <erlehmann> | AryehGregor, how about a mail system via XMPP? that would be rad. |
| 20:26 | <Hixie> | TabAtkins: huh, i didn't know you could do that |
| 20:26 | <AryehGregor> | Hixie, sure you can. |
| 20:26 | <Hixie> | interesting |
| 20:26 | Hixie | pokes |
| 20:27 | <AryehGregor> | erlehmann, how about any mail system that actually implements meaningful authentication but is still backward-compatible with SMTP? |
| 20:27 | <AryehGregor> | You come up with that and I'm sold. |
| 20:27 | <erlehmann> | AryehGregor, i reject your compatibility concerns and substitute my own. |
| 20:27 | <AryehGregor> | Oh, and that's usable by normal people, so it can eventually supplant SMTP. |
| 20:27 | <zewt> | years ago I tried the "separate email aliases for everything" thing. all it did was make me receive every spam 40x, once to each, heh |
| 20:27 | <erlehmann> | claws-mail actually treats RSS and ATOM feeds like incoming mail. that is a pretty interesting idea. |
| 20:27 | <AryehGregor> | Mailing lists are a big problem. |
| 20:28 | <Hixie> | the thing i love about smtp is the ability to send mail from a script with an arbitrary from: address |
| 20:28 | <zewt> | handy in principle for tracking down sites who sell email addresses, but that only actually happened maybe once or twice |
| 20:28 | <Hixie> | this is also the thing that allows the spam problem to exist |
| 20:28 | <erlehmann> | i did that, once. i was like 12 and it was like telnet. |
| 20:28 | <AryehGregor> | zewt, yeah, that's largely been my experience too. |
| 20:28 | <annevk> | Hixie, I think so |
| 20:28 | <AryehGregor> | I continue to do it, but mostly out of habit. |
| 20:28 | <Hixie> | annevk: k |
| 20:29 | <annevk> | Hixie, actually, apart from XHR, I sort of thought we'd do it that way for all cross-origin resources |
| 20:29 | <AryehGregor> | I should probably just switch everything to using ayg⊙an and not worry about spambots getting it. |
| 20:29 | <annevk> | Hixie, avoiding the need for crossorigin="" |
| 20:29 | <erlehmann> | are arbitrary from addresses supported in MUAs besides claws-mail? |
| 20:29 | <zewt> | annevk: also, welcome back :P |
| 20:29 | <annevk> | hey zewt |
| 20:29 | <erlehmann> | just report the spammers to cyber police and state police. consequences will never be the same. |
| 20:29 | <AryehGregor> | erlehmann, IIRC, GNU mail supports it, but only if you're root. |
| 20:29 | <Hixie> | annevk: for <img> we figured that static images would often need cross-origin support and having the server reflect the origin each time was going to be a huge pain |
| 20:29 | <AryehGregor> | Not that non-root can't just use telnet. |
| 20:30 | <AryehGregor> | I've done it via telnet more than once. |
| 20:30 | <Hixie> | annevk: so we had to have a way that wouldn't send the cookies so the site could use "8" |
| 20:30 | <Hixie> | er |
| 20:30 | <Hixie> | "*" |
| 20:30 | <erlehmann> | AryehGregor, >implying i'd use any program as root. |
| 20:30 | <AryehGregor> | erlehmann, do you run init as root? :) |
| 20:30 | <Ms2ger> | AryehGregor, does krijnh strip email addresses from the logs? |
| 20:30 | <erlehmann> | AryehGregor, core utils et al are exempt! |
| 20:31 | <AryehGregor> | Ms2ger, I dunno, but to preserve my sanity, I assume spam harvesters don't actually read every single page on the web. |
| 20:31 | <Hixie> | AryehGregor: they definitely read w3.org. hedral⊙dc was on w3.org for like 5 seconds before it started gettign tons of spam. |
| 20:31 | <erlehmann> | AryehGregor, do you make the irc log page? |
| 20:32 | <AryehGregor> | erlehmann, do you start webservers as root? |
| 20:32 | <AryehGregor> | erlehmann, krijn makes the IRC log page. |
| 20:32 | <annevk> | Hixie, ah yeah, I think we should support * for credentialed requests |
| 20:33 | <Hixie> | annevk: well for eventsource it doesn't matter, they have to have a script there anyway or the feature is pointless |
| 20:33 | <erlehmann> | AryehGregor, i test software locally on port 8080. but i see what you did there, YOU WIN. enjoy your cake. |
| 20:33 | <AryehGregor> | :) |
| 20:34 | <erlehmann> | that reminds me, i need to read CSSquirrel. the google-GLaDOS really cracks me up. |
| 20:38 | <annevk> | Hixie, yeah |
| 20:42 | <Hixie> | anne: man, what a tangled web we weave |
| 20:42 | <Hixie> | annevk: event source calls one of my algorithms which invokes cors which invokes fetch which invokes http |
| 20:43 | <Hixie> | annevk: at some point when we've both got some free time we should work out how to shuffle things around cors, xhr, html, and dom core, and rationalise these dependencies |
| 20:43 | <Hixie> | annevk: (no rush on that though... maybe next year) |
| 20:44 | <annevk> | yeah, CORS is kind of awkward |
| 20:45 | <annevk> | I think DOM Core is alright as low-level dependency though |
| 20:45 | <Ms2ger> | Now you just need to remove that stuff from HTML |
| 20:51 | <Hixie> | Ms2ger, annevk: before doing that we need to propagate recent changes to html |
| 20:51 | <Hixie> | Ms2ger, annevk: it's on my list, but not a high priority... in the meantime i'm kinda worried the text is forking |
| 20:51 | <Ms2ger> | I'm trying to follow your changes |
| 20:52 | <Hixie> | k |
| 20:53 | <Hixie> | i can put markers in the .../source to make it trivial for you to just slurp it in if you prefer |
| 20:54 | <Ms2ger> | No need, it's not like it changes often |
| 20:55 | <Hixie> | k |
| 20:57 | <Hixie> | there's really no good api for the postMessage() transfer thing that i can see |
| 20:57 | <Hixie> | all the options suck in some way |
| 20:58 | <Hixie> | dunno what to do about that |
| 20:59 | <zewt> | the main proposal right now doesn't seem to suck very much |
| 21:02 | <annevk> | doesn't sound like it's great either :) |
| 21:03 | <zewt> | don't think there are any serious problems with it |
| 21:03 | <Hixie> | zewt: which is the "main" one? |
| 21:03 | <zewt> | postMessage([obj1, obj2, obj3], [obj2]) |
| 21:03 | <Hixie> | and what do you get on the other side? |
| 21:03 | <zewt> | [obj1, obj2, obj3] |
| 21:04 | <Hixie> | what happens if something is only in the first object or only in the second array? |
| 21:04 | <zewt> | one sec I think I summarized how I was thinking of it in a mail, let me find it |
| 21:05 | <Hixie> | k, i'll check it when i get back |
| 21:05 | <Hixie> | lunch now |
| 21:05 | <Hixie> | bbiab |
| 21:07 | <zewt> | http://lists.w3.org/Archives/Public/public-webapps/2011AprJun/0977.html |
| 21:11 | The_8472 | wonders if webworkers are actually good for anything but numbercrunching |
| 21:12 | <TabAtkins> | Everything is number-crunching, so what's the difference? |
| 21:13 | <The_8472> | interaction with the API |
| 21:13 | <zewt> | well, not really (not by what the term typically means), but there are plenty of non-number-crunching use cases: http://www.whatwg.org/specs/web-apps/current-work/complete/workers.html |
| 21:13 | <The_8472> | with any API |
| 21:14 | <The_8472> | 'In this example, the main document spawns a worker whose only task is to listen for notifications from the server, and, when appropriate, either add or remove data from the client-side database.' <- that could be done with an event based API |
| 21:14 | <The_8472> | instead of blocking for some notification |
| 21:14 | <zewt> | so? |
| 21:14 | <The_8472> | no need for an extra thread |
| 21:14 | <zewt> | a huge set of use cases for threads is turning ugly event-based code into clean linear code |
| 21:14 | <charlvn> | a lot of dom work can also bog down slower processors |
| 21:15 | <zewt> | far more common than number crunching, in my experience (for threads in all environments, not just workers) |
| 21:15 | <The_8472> | dom work can't be webworker'd, can it? |
| 21:15 | <charlvn> | ouch |
| 21:16 | <The_8472> | the problem with dom and javascript is that they haven't been designed for multithreading |
| 21:16 | <The_8472> | so we have practically no thread-safe APIs |
| 21:16 | <The_8472> | which puts strong limits on what can be passed between threads |
| 21:16 | <zewt> | also, "no need for an extra thread" doesn't matter--threads aren't expensive, there's no reason to *avoid* the extra thread |
| 21:16 | <charlvn> | yeah i see, that is a problem |
| 21:17 | <The_8472> | there is when the thread-communication incurs is unflexible |
| 21:17 | <The_8472> | err |
| 21:17 | <charlvn> | in the examples the dom work is being done using the onmessage callback |
| 21:17 | <charlvn> | so i guess that will run in the main thread |
| 21:17 | <The_8472> | yeah |
| 21:18 | <charlvn> | but i think that's cool though, it's cleaner |
| 21:18 | <charlvn> | and besides, browsers should opmise that type of stuff in the background |
| 21:18 | <zewt> | you should really be able to do canvas/img/webgl stuff in workers, but hopefully that'll come eventually |
| 21:19 | <charlvn> | that would be a necessity i think |
| 21:19 | <The_8472> | that's the point. we need thread-safe APIs. or at least a way to pass native object ownership to another thread |
| 21:19 | <erlehmann> | zewt, then javascript can mine bitcoins even better! |
| 21:19 | <zewt> | well, there's work on object transfer right now, which is the framework needed to support transferring more complex objects (though don't hold your breath on it being used for that any time soon) |
| 21:24 | <The_8472> | <erlehmann> zewt, then javascript can mine bitcoins even better! <- that doesn't really require many threads. it needs access to shaders |
| 21:24 | <erlehmann> | hehe |
| 21:25 | <sicking> | Hixie: ping |
| 21:35 | <zcorpan> | rwaldron: i'm here now |
| 21:36 | <rwaldron> | zcorpan good afternoon! |
| 21:36 | <zcorpan> | afternoon! |
| 21:36 | <rwaldron> | I'm not sure how familar you are with Popcorn.js |
| 21:37 | <rwaldron> | it is a Mozilla project that I'm involved in whose goal is to add a programmatic layer to html5 video |
| 21:37 | <rwaldron> | really fun stuff |
| 21:38 | <zcorpan> | hadn't heard of it |
| 21:38 | <rwaldron> | anyway, it has come to light that the wording of the "canplaythrough" event's preconditions have been interpretted differently by firefox engineers and chrome engineers |
| 21:39 | <rwaldron> | firefox fires the event whenever readyState is HAVE_ENOUGH_DATA, which is variant |
| 21:39 | <zcorpan> | and differently again by opera engineers (though that's a bug) :) |
| 21:39 | <rwaldron> | chrome is firing it once, at the beginning, when this first happens |
| 21:39 | <rwaldron> | it seems that the firefox interpretation is correct |
| 21:39 | <rwaldron> | however this leaves a gap |
| 21:40 | <rwaldron> | essentially, if i attach an event listener to a media element for canplaythrough, that modifies the currentTime, it will fire the event again |
| 21:40 | <rwaldron> | ...and of course... again. |
| 21:40 | <rwaldron> | forever. |
| 21:40 | <zcorpan> | from the spec it seems firefox is correct in that case |
| 21:41 | <rwaldron> | yes, agreed |
| 21:41 | <zcorpan> | but it could be an unintended spec bug |
| 21:41 | <zcorpan> | canplay has a condition that canplaythrough doesn't have |
| 21:41 | <rwaldron> | so we discussed earlier in irc.mozilla.org/popcorn-js the need for an event that first once, but with the same precondition as "canplaythrough" |
| 21:41 | <rwaldron> | i wrote up the event details: https://gist.github.com/1031549 |
| 21:42 | <rwaldron> | canplay's preconditions is HAVE_FUTURE_DATA |
| 21:42 | <zcorpan> | rwaldron: sounds like you're reading a non-normative table in the spec |
| 21:42 | <rwaldron> | http://www.w3.org/TR/html5/video.html#event-media-canplay |
| 21:43 | <rwaldron> | indeed i am |
| 21:43 | <zcorpan> | you're also reading an out of date version of the spec |
| 21:43 | <rwaldron> | i'm open to guidance |
| 21:43 | <zcorpan> | http://www.whatwg.org/specs/web-apps/current-work/complete/the-iframe-element.html#the-video-element <- read this document |
| 21:44 | <zcorpan> | search for "canplay" or "canplaythrough" and make sure the section you're in isn't marked as "non-normative" |
| 21:44 | <rwaldron> | so, they are essentially the same, correct? |
| 21:44 | <zcorpan> | yeah |
| 21:44 | <zcorpan> | but in general you should never read TR/ |
| 21:44 | <rwaldron> | ok, good to know |
| 21:45 | <zcorpan> | since it's actually out-of-date-snapshot-with-issues-that-might-be-fixed-in-the-editor's-draft/ |
| 21:45 | <annevk> | TR/ is where you go if you want to implement twice |
| 21:45 | <rwaldron> | the only place i see these events listed with details about preconditions is in the table marked non-normative |
| 21:45 | <annevk> | or more dramatic, it is where implementors go to die |
| 21:45 | <rwaldron> | gotcha |
| 21:46 | <zcorpan> | yeah the preconditions you need to figure out yourself by reading the algorithms where it says to fire the events |
| 21:47 | <rwaldron> | right, those match (or, I am interpretting them to match) the preconditions listed in the table |
| 21:47 | <zcorpan> | "When the ready state of a media element whose networkState is not NETWORK_EMPTY changes, the user agent must follow the steps given below:" |
| 21:47 | <rwaldron> | right, got it |
| 21:48 | <rwaldron> | so basically, we implemented an events system for delegating nth queued callbacks to specified events |
| 21:48 | <rwaldron> | as of today we've had to create a "custom" event that behaves in a way that is described in the gist |
| 21:49 | <rwaldron> | which is why I'm reaching out |
| 21:49 | <zcorpan> | i don't follow |
| 21:50 | <rwaldron> | the need for a _single_ event to fire when readyState has newly reached HAVE_ENOUGH_DATA, but only the first time |
| 21:50 | <zcorpan> | ah |
| 21:50 | <rwaldron> | basically, the "polyfill" attaches a handler that listens for "canplaythrough" event, but supresses all future dispatches of the event |
| 21:50 | <zcorpan> | i wonder if opera does it like firefox or like chrome |
| 21:51 | <rwaldron> | not sure |
| 21:51 | <zcorpan> | rwaldron: we could change the spec here |
| 21:51 | <zcorpan> | i don't see why it's useful to fire canplaythrough several times |
| 21:51 | <rwaldron> | nor do i |
| 21:51 | <rwaldron> | too be honest |
| 21:51 | <rwaldron> | we've been interpretting it as a single event all along |
| 21:51 | <rwaldron> | of course, incorrectly |
| 21:53 | <zcorpan> | could you file a spec bug asking for canplay and canplaythrough to never be fired twice? |
| 21:53 | <rwaldron> | i certainly can |
| 21:53 | <rwaldron> | where should I file? |
| 21:53 | <zcorpan> | use the comment box at the bottom of the spec |
| 21:54 | <rwaldron> | ok, awesome, thats what i figured, just wanted to make sure |
| 21:54 | <humph> | zcorpan: the reason firefox does it, or so I was told, is because of seeks and starting at new offsets into the media resource |
| 21:54 | <zcorpan> | prefix it with "<video>" |
| 21:56 | <zcorpan> | humph: the use case for canplaythrough i think is for when to start playing the video |
| 21:56 | <humph> | I agree |
| 21:56 | <Hixie> | sicking: here |
| 21:56 | <zcorpan> | not sure what canplaythrough would be useful for after a seek |
| 21:56 | <humph> | I'm relaying what cpearce and kinetik were telling me |
| 21:56 | <humph> | agreed |
| 21:56 | <zcorpan> | k |
| 21:56 | <humph> | they didn't agree with me, though |
| 21:57 | <zcorpan> | Hixie: you happen to know the design decisions around whether canplaythrough should fire more than once per spec? |
| 21:57 | <Hixie> | shouldn't it play whenever the UA transitions from can't play through to can play through? |
| 21:58 | <Hixie> | zewt: so what happens if something is only in the second argument? |
| 21:58 | <rwaldron> | zcorpan is this good: |
| 21:58 | <rwaldron> | "<video> canplay and canplaythrough should only fire once, when the new readyState is HAVE_FUTURE_DATA and HAVE_ENOUGH_DATA for the first time, respectively." |
| 21:58 | <humph> | zcorpan: https://bugzilla.mozilla.org/show_bug.cgi?id=664842#c1 |
| 21:58 | <zcorpan> | rwaldron: yeah. rationale would be good too but can be put in a comment of the bug |
| 21:59 | <zewt> | Hixie: my first impression is nothing, unless it's really worth tracking which were used and throwing an error |
| 21:59 | <rwaldron> | ok |
| 21:59 | <zewt> | the second argument just saying "if you see one of these, enable its transfer behavior" |
| 22:00 | <Hixie> | zewt: so if all you want to do is pass a port you have to list it twice? that's pretty lame. |
| 22:00 | <rwaldron> | zcorpan humph sdowne http://www.w3.org/Bugs/Public/show_bug.cgi?id=12982 |
| 22:00 | <zewt> | minorly lame |
| 22:00 | <zcorpan> | Hixie: yes, point is that that can happen several times if you seek |
| 22:00 | <Hixie> | zewt: as minorly as all the other proposals |
| 22:00 | <Hixie> | zcorpan: yes |
| 22:00 | <zewt> | i don't recall seeing any other proposals that didn't have major problems (usually forwards-compatibility) |
| 22:01 | <zcorpan> | Hixie: is that intentional? |
| 22:02 | <Hixie> | zcorpan: yes, but it can change if there's good reason to |
| 22:02 | <Hixie> | zewt: how about postMessage({object to clone}, [objects to transfer]); ? |
| 22:03 | <Hixie> | zewt: with the other side receiving a clone of the object to clone, and an array of the transferred objects separately |
| 22:03 | <zcorpan> | Hixie: rwaldron writes a html5 media js framework and needs to work around that it can fire more than once |
| 22:03 | <zewt> | i think that's very seriously lame: you have to remove the objects you want to transfer from the object tree, then put them back in on the other side |
| 22:03 | <Hixie> | zcorpan: why? what does he do on that event that isn't idempotent? |
| 22:04 | <rwaldron> | Hixie, this is pasted from earlier: essentially, if i attach an event listener to a media element for canplaythrough, that modifies the currentTime, it will fire the event again |
| 22:04 | <Hixie> | zewt: no, we can just say that if it's in both, it gets transferred, as with yours |
| 22:04 | <zewt> | it also means you can't trivially enable transfer for an object if the UA supports it, falling back on cloning if not |
| 22:04 | <rwaldron> | and that will cause a loop of the "canplaythrough" event |
| 22:04 | <zewt> | i guess |
| 22:04 | <Hixie> | rwaldron: why would you do that? |
| 22:04 | <zewt> | personally I'm very ambivalent about the "deprecate the ports list" thing, since I'm guessing it's already been around long enough that it'll never actually go away |
| 22:04 | <rwaldron> | zcorpan Hixie - i am joined in that effort by humph and sdowne who are also present |
| 22:05 | <Hixie> | rwaldron: just seek as soon as you can, not just once you have some random subset of the video buffered |
| 22:05 | <rwaldron> | Hixie "canplaythrough" is the only event that signifies HAVE_ENOUGH_DATA |
| 22:05 | <zewt> | so in principle, the ports list could become a "list of transferred objects" |
| 22:05 | <Hixie> | rwaldron: if you're going to seek, HAVE_ENOUGH_DATA is uninteresting |
| 22:05 | <Hixie> | rwaldron: as soon as you seek, all bets are off as to whether you still have any data at all |
| 22:06 | <rwaldron> | sdowne humph, can you guys explain the needs that arose from that test case? |
| 22:06 | <humph> | I think it's not something that can't be overcome. it came out of a reading of the spec that assumed a one time event |
| 22:06 | <humph> | that it fires more than once is fine, but confusing given that impls differ so much |
| 22:06 | <zewt> | Hixie: what happens if an object is in the transfer list which the browser doesn't have transfer support for? |
| 22:06 | <rwaldron> | we have in fact already written a workaround that gives us what we need |
| 22:07 | <Hixie> | yeah, the browsers differing is a real problem |
| 22:07 | <zcorpan> | foolip: you happen to know whether opera can fire canplaythrough more than once? |
| 22:08 | <zewt> | give null on the receiving side? |
| 22:08 | <Hixie> | zewt: it would throw, like with your proposal, because the IDL of the second argument would be Transferable[] so you'd get a TypeError |
| 22:09 | <zewt> | then you'd have to carefully ensure that you test transferrability for each object before you put it in the array, which is sort of inconvenient |
| 22:10 | <sicking> | Hixie: what spec defines all the new media streams stuff? I'm asking in particular since i'd like for media streams to use a interface not named simply Stream as I'd like for us to use that for data streams where you can actually get at the binary data |
| 22:10 | <Hixie> | zewt: what would you suggest? |
| 22:10 | <Hixie> | sicking: HTML |
| 22:11 | <zewt> | the list being a request, not a requirement: if transfer for an object isn't supported, fall back on a regular clone |
| 22:12 | <sicking> | Hixie: whereas as I understand it, people aren't really expected to read the raw data from media streams, but rather forward them to things which operate on the pixel data, or display them in a <video>/<audio> |
| 22:12 | <zewt> | so you can say postMessage([obj1, obj2, obj3], [obj1, obj3]), and if the browser only supports transfer for obj3, that's what happens |
| 22:13 | <sicking> | Hixie: where should i file a bug to get this changed? Or send an email |
| 22:13 | <zewt> | otherwise, you'd write that code in a new browser and it'd work, but it would fail on older browsers only supporting transfers for obj1, which doesn't seem good for backwards-compat |
| 22:14 | <zcorpan> | rwaldron: if you can come up with reasons why firing it multiple times is bad/confusing/wrong, comment on the bug to convince Hixie :) |
| 22:14 | <AryehGregor> | sicking, same place as other HTML stuff, either an e-mail to whatwg or a bug in the W3C tracker. There's a component that can be used for stuff not in the W3C draft. |
| 22:14 | <rwaldron> | zcorpan ok |
| 22:14 | <zcorpan> | rwaldron: otherwise he'll just go "meh" and WONTFIX it |
| 22:15 | <sicking> | AryehGregor: Is that the HTML.next product? |
| 22:15 | <AryehGregor> | sicking, it really doesn't matter, as long as it's something with Hixie as the editor. He treats them all interchangeably. |
| 22:15 | <AryehGregor> | I think "other Hixie drafts" would be fine. |
| 22:16 | <sicking> | i'll send to the whatwg list, less politics |
| 22:16 | <AryehGregor> | In the HTML WG product. |
| 22:16 | <AryehGregor> | :) |
| 22:16 | <Hixie> | sicking: i can just rename it now if you want |
| 22:16 | <sicking> | Hixie: that would be lovely :) |
| 22:16 | <Hixie> | sicking: (sorry for delay, had something come up) |
| 22:16 | <zcorpan> | MediaStream? |
| 22:16 | <sicking> | yes! |
| 22:17 | <Hixie> | would it inherit from Stream? |
| 22:17 | <Hixie> | Generally FooBar is an object that inherits from Bar, on the Web |
| 22:17 | <sicking> | that'd defaeat the purpose I think |
| 22:17 | <sicking> | defeat even |
| 22:17 | <sicking> | what I'd like to avoid is to have to define what encoding you get if you read the raw bytes |
| 22:17 | <zcorpan> | Hixie: there are exceptions |
| 22:17 | <Hixie> | how would one get ahold of these binary streams? |
| 22:17 | <zcorpan> | EventSource |
| 22:17 | <zcorpan> | WebSocket |
| 22:17 | <Hixie> | zcorpan: there's no Source or Socket |
| 22:17 | <sicking> | or what happens if you plug a binary-data stream into something that operates on pixels |
| 22:18 | <zcorpan> | exactly |
| 22:18 | <sicking> | Hixie: i suspect XHR should be able to return a stream eventually |
| 22:18 | <Hixie> | zcorpan: there would be a Stream in this proposal |
| 22:18 | <Hixie> | sicking: ah, interesting |
| 22:18 | <sicking> | Hixie: xhr.responseType = "stream" |
| 22:18 | <Hixie> | makes sense |
| 22:18 | Hixie | opens his thesaurus |
| 22:18 | <sicking> | Hixie: likely we'll have encoders/decoders that can convert between MediaStreams and Streams |
| 22:18 | <zcorpan> | Hixie: point |
| 22:19 | <Hixie> | sicking: yeah |
| 22:19 | <sicking> | Hixie: where you can do things like choose quality and algorithm |
| 22:20 | <Hixie> | hmm |
| 22:20 | <Hixie> | MediaStream would be good except for the clash with Stream later |
| 22:20 | <sicking> | possibly it'd make sense to hook up a stream to a specific WebSocket message (both sending and receiving) |
| 22:21 | Hixie | looks at the spec to see how else one could describe these objects |
| 22:22 | <sicking> | Hixie: I'm not *hugely* concerned about the name conflict (if that's the only conflict there is). But of course if we can find something better then that's great too |
| 22:24 | <zcorpan> | MediaBlow |
| 22:24 | <zcorpan> | MediaOoze |
| 22:25 | <zcorpan> | MediaRainCatsAndDogs |
| 22:25 | zcorpan | closes the synonym dictionary |
| 22:28 | <rwaldron> | zcorpan i've included a description of how this issue can actually lock up media elements.. fairly compelling |
| 22:28 | <rwaldron> | also, link to a demo |
| 22:28 | <zcorpan> | awesome |
| 22:28 | <mpilgrim> | webkit's IDL bindings are finally getting more strict |
| 22:29 | <rwaldron> | that show canplaythrough firing infinitely |
| 22:29 | <rwaldron> | the page is currently at >60k canplaythrough events |
| 22:29 | <mpilgrim> | er, that was @sicking: webkit's IDL bindings are finally getting more strict |
| 22:29 | <rwaldron> | and the video is unplayable |
| 22:29 | <rwaldron> | as i collect evidence, i will post |
| 22:31 | <mpilgrim> | sicking: soon all IndexedDB methods will throw on missing required arguments |
| 22:31 | <sicking> | mpilgrim: which ones weren't strict before? |
| 22:32 | <sicking> | mpilgrim: you mean in spec or in implementations? |
| 22:32 | <mpilgrim> | virtually all of them |
| 22:32 | <mpilgrim> | i mean webkit's implementation |
| 22:32 | <sicking> | mpilgrim: ah, excellent |
| 22:32 | <Hixie> | sicking: let me poke at this interface name some more, if you don't see it change by tonight then please chase me down with pitchforks |
| 22:33 | <mpilgrim> | webkitIndexedDB.open() will throw, for example |
| 22:33 | <sicking> | Hixie: one option is to have MediaStream for what you have now, and DataStream for the other thing i'm talking about |
| 22:33 | <Hixie> | sicking: yeah, i was thinking maybe BlobStream for the other thing |
| 22:33 | <rwaldron> | zcorpan care to dance over another media element zinger? |
| 22:33 | <rwaldron> | :D |
| 22:33 | <zcorpan> | sure |
| 22:33 | <sicking> | Hixie: we can bikeshed on that later ;-) |
| 22:34 | <Hixie> | sicking: :-) |
| 22:34 | <rwaldron> | this one is awesome |
| 22:34 | <rwaldron> | http://www.w3.org/Bugs/Public/show_bug.cgi?id=12541 |
| 22:34 | <rwaldron> | i filed sometime ago |
| 22:35 | <Hixie> | hm, should GeneratedStream.stop() be .close()? that might make more sense. |
| 22:35 | <rwaldron> | my investigation shows that _every_ UA is doing something in order to make up for this |
| 22:36 | <rwaldron> | it was initially discovered when i was trying to write a method for the popcorn.js lib that returned a rounded currentTime |
| 22:36 | <rwaldron> | so, i wrote a failing unit test |
| 22:36 | <zcorpan> | rwaldron: like foolip said in the bug, that's because the spec used to say something different and browsers haven't caught up yet |
| 22:36 | <rwaldron> | that set the current time to 0.89 |
| 22:36 | <rwaldron> | from 0 |
| 22:36 | <rwaldron> | and roundTime was expected to return 1 |
| 22:37 | <rwaldron> | hrm |
| 22:37 | zcorpan | reads the spec |
| 22:37 | rwaldron | looking that the spec you linked to me earlier |
| 22:37 | <Hixie> | maybe GeneratedStream should be LocalMediaStream and Stream should just be MediaStream |
| 22:38 | <rwaldron> | zcorpan still the same |
| 22:38 | <rwaldron> | http://www.whatwg.org/specs/web-apps/current-work/complete/the-iframe-element.html#seeking |
| 22:38 | <rwaldron> | there is a misunderstanding somewhere |
| 22:39 | <rwaldron> | basically, if the async steps dont happen fast enough, after setting currentTime, then getting currentTime will return the previous time, not the new time i just set to |
| 22:40 | <rwaldron> | which makes it impossible to test if a media element is at the right place in time without writing some kind of async deferred nonsense to return the correct time whenever the UA gets around to it. |
| 22:40 | <zcorpan> | yeah that's my reading as well |
| 22:42 | <rwaldron> | so, i had filed this in the firefox bugzilla and despite initial objections (because, yes - they were writing to spec) was brought into parity with the other UAs |
| 22:42 | <rwaldron> | which were all doing some kind of cached last currentTime calculation |
| 22:42 | <rwaldron> | /rant |
| 22:42 | <rwaldron> | :) |
| 22:43 | <rwaldron> | I just want to say that I really appreciate your time and attention this afternoon |
| 22:44 | <zcorpan> | commented on the bug |
| 22:45 | <Hixie> | anyone remember why i added videoTracks and audioTracks to GeneratedMediaStream? |
| 22:45 | <Hixie> | er |
| 22:45 | <Hixie> | LocalMediaStream, previously known as GeneratedStream |
| 22:46 | <Hixie> | (added in r5965 2011-03-25) |
| 22:46 | <TabAtkins> | So you could get access to the video and audio tracks more easily? |
| 22:46 | <Hixie> | when would there be more than one? |
| 22:46 | <TabAtkins> | I think videos can have multiple of each. |
| 22:47 | <Hixie> | not those from the local camera though... |
| 22:48 | <Hixie> | aha http://lists.whatwg.org/htdig.cgi/whatwg-whatwg.org/2011-March/031058.html |
| 22:48 | <rwaldron> | zcorpan awesome :) |
| 22:48 | <Hixie> | oh wow, i suck, that totally doesn't address that use case |
| 22:50 | <Hixie> | we still need some way to pause outgoing video without affecting the LocalMedisStream or the local <video> mirror. |
| 22:50 | <Hixie> | hmm |
| 22:53 | <zcorpan> | rwaldron: got any others before i go? :) |
| 22:53 | <rwaldron> | nope! |
| 22:53 | <rwaldron> | but, again... thank you |
| 22:53 | <zcorpan> | bummer |
| 22:54 | <zcorpan> | np |
| 22:54 | <rwaldron> | if you didnt see above |
| 22:54 | <rwaldron> | you're time is greatly appreciated |
| 22:54 | <zcorpan> | anytime |
| 22:54 | <rwaldron> | i work on jQuery and I know how volunteer stuff goes |
| 22:54 | <rwaldron> | so i want to convey my appreciation |
| 22:54 | <rwaldron> | have a good night :D |
| 22:54 | <zcorpan> | likewise :) |
| 22:58 | <zcorpan> | Specification: |
| 22:58 | <zcorpan> | http://www.whatwg.org/specs/web-apps/current-work/complete/common-microsyntaxes.html |
| 22:58 | <zcorpan> | Section: http://www.whatwg.org/specs/web-apps/current-work/#colors |
| 22:59 | <zcorpan> | Hixie: could you please make the comment script emit http://www.whatwg.org/specs/web-apps/current-work/complete/common-microsyntaxes.html#colors instead of the above? |
| 23:00 | <Hixie> | zcorpan: the first link is the referer |
| 23:00 | <Hixie> | zcorpan: the second link is for me and is to the HTML spec or the complete.html spec with the appropriate fragid |
| 23:00 | <zcorpan> | oh |
| 23:00 | <Hixie> | zcorpan: though i might change it to just be complete.html always, since i've basically stopped using the HTML one |
| 23:01 | <zcorpan> | Hixie: could you add the fragment to the first url? |
| 23:04 | <Hixie> | what for? |
| 23:04 | <zcorpan> | i don't want to load the full spec so i need to copy-paste the fragment from the second url, click on the first one adn then paste in the fragment again |
| 23:05 | <zcorpan> | which is annoying |
| 23:05 | <zcorpan> | s/-paste// |
| 23:05 | <Hixie> | the first url is going to be the full spec often too |
| 23:05 | <Hixie> | if you just want a short-spec link version isntead i can do that, though |
| 23:06 | <zcorpan> | yeah that'd be nice |
| 23:06 | <AryehGregor> | Yeah, please have a link to the multipage version somewhere. |
| 23:06 | <AryehGregor> | With the right fragment. |
| 23:06 | <AryehGregor> | The status quo is annoying in that respect. |
| 23:06 | <Hixie> | it'll be the one with the redirects, since i have no idea hwo to generate the full ones, is that ok? |
| 23:06 | <zcorpan> | that's fine |
| 23:07 | <zcorpan> | complete/ is fine |
| 23:07 | <Hixie> | done |
| 23:07 | <Hixie> | (untested) |
| 23:07 | <zcorpan> | thanks |
| 23:07 | <zcorpan> | everyone! file a test spam bug! |
| 23:07 | <Hixie> | nooo |
| 23:08 | <Hixie> | just find a real bug then file that! |
| 23:08 | <Hixie> | actually i have a bug to file, let me test it now |
| 23:10 | <Hixie> | ironically, the urls in question are bogus already |
| 23:12 | <zcorpan> | what's bogus? |
| 23:14 | <Hixie> | zcorpan: in the bug i just filed, the url was #generatedstream |
| 23:15 | <zcorpan> | oh and you renamed it? |
| 23:16 | <Hixie> | yeah |
| 23:16 | <Hixie> | Stream, GeneratedStream, and StreamRecorder became MediaStream, LocalMediaStream, and MediaStreamRecorder |
| 23:16 | <Hixie> | sicking: ^ |
| 23:17 | <zcorpan> | nn |
| 23:17 | <Hixie> | nn |
| 23:18 | <sicking> | Hixie: sweet! What format does MediaStreamRecorder record things in? |
| 23:19 | <sicking> | Hixie: and if it's an encoded format, can you set encoding parameters such as quality? |
| 23:19 | <sicking> | Hixie: to be clear, I think this is strictly an improvement. So absolutely a good thing. These questions applied before too |
| 23:19 | <sicking> | Hixie: so thanks :) |
| 23:19 | <Hixie> | that's as yet undefined, other than it must be a format supported by the local <video>. Same problem with getObjectURL() for a MediaStream and for sending data via a PeerConnection object. |
| 23:20 | <Hixie> | i'm waiting implementation experience before specifying the api for deciding the encoding parameters |
| 23:20 | <Hixie> | and i'm waiting for a miracle before specifying the codec |
| 23:20 | <sicking> | haha |
| 23:20 | <sicking> | sounds good |
| 23:21 | <sicking> | i'll be doing the swedish encoder dance to appease the encoder gods |
| 23:21 | <Hixie> | man at this point i'm willing to try anything |
| 23:21 | <sicking> | i'll fax you the steps |
| 23:22 | smaug____ | missed all the context, but he assumes sicking has seen roc's MediaStream API |
| 23:23 | <sicking> | smaug____: i have |
| 23:23 | <roc> | ah good, renaming to MediaStream was something I was going to ask for |
| 23:23 | <rwaldron> | Hixie thanks for responding to the "canplaythrough" ticket so quickly |
| 23:24 | <rwaldron> | damn shame that it was stomped on by a mozilla employee |
| 23:24 | <rwaldron> | though it was discovered in a mozilla project |
| 23:24 | <rwaldron> | and honestly... |
| 23:24 | <Hixie> | rwaldron: pure luck in this particular case, normally i go in date order but i happened to be looking at recent bugs for the cgi thing zcorpan and AryehGregor were asking me about |
| 23:25 | <Hixie> | rwaldron: feel free to reopen if it needs more thought :-) |
| 23:25 | <rwaldron> | a HAVE_ENOUGH_DATA canplaythrough that fires many times is useful, but there is still no event that signifies this has happened when it happens for the first time |
| 23:25 | <rwaldron> | anyway, we've written a fallback that does what we need. so its all good |
| 23:25 | <cpearce> | rwaldron: that's pretty easy to track yourself if it matters. |
| 23:25 | <Hixie> | you can tell it happened for the first time by just keeping track of whether you've seen it or not |
| 23:25 | <kinetik> | rwaldron: "stomped on" is not a fair categorization. |
| 23:25 | <rwaldron> | cpearce yeah, aware of that and already doing so |
| 23:26 | <rwaldron> | yes. Hixie that is in fact what we're doing |
| 23:26 | <Hixie> | cool |
| 23:26 | <rwaldron> | though I'm sure that the test of time will show more then just a handful of developers being confused by this behaviour |
| 23:27 | <rwaldron> | also, keep in mind that silvia pfieffer's book actually says this fires "for the first time" |
| 23:27 | <rwaldron> | so yeah thats newly in print |
| 23:27 | <rwaldron> | anyone reading this will also be mislead |
| 23:27 | <rwaldron> | kinetik i'd say is. |
| 23:27 | <rwaldron> | especially considering your choice of words regarding the test case |
| 23:27 | <cpearce> | only on page 111, not on page 129. ;) |
| 23:28 | <roc> | the spec trumps the book :-) |
| 23:28 | <rwaldron> | which is perfectly valid if you you want to jump to a specific time in the media |
| 23:28 | <rwaldron> | roc i get that |
| 23:28 | <rwaldron> | but think about the average "web designer" |
| 23:28 | <roc> | I am sad that three of us are on IRC on a Saturday morning |
| 23:28 | <rwaldron> | do they read the spec? |
| 23:28 | <cpearce> | :P |
| 23:28 | <roc> | rwaldron: I do, a lot |
| 23:29 | <rwaldron> | i'd say an exception tbh |
| 23:29 | <roc> | rwaldron: if you want to help the average web developer, please lobby for bug 12267 :-) |
| 23:29 | <roc> | rwaldron: they don't, but it's incumbent on book authors to follow the spec, not the other way around |
| 23:29 | <roc> | there will be many books |
| 23:29 | <roc> | as the spec changes, books will need to be revised and updated |
| 23:30 | <rwaldron> | roc, to be clear, I completely agree |
| 23:30 | <rwaldron> | here's where I'm coming from... |
| 23:30 | <rwaldron> | I'm on the jQuery bugs team |
| 23:31 | <rwaldron> | we field ~50 tickets per day |
| 23:31 | <rwaldron> | have you seen what the lowest common denominator is writing for code? |
| 23:31 | <roc> | yes |
| 23:32 | <roc> | I have been reading Firefox bug reports for over ten years :-) |
| 23:32 | <rwaldron> | cool, so you see where I'm coming from |
| 23:32 | <roc> | OK that's a lie |
| 23:32 | <roc> | the first few years were Netscape bug reports :-) |
| 23:32 | <rwaldron> | hahaha well... then you def see where i'm coming from |
| 23:32 | <rwaldron> | :) |
| 23:32 | <rwaldron> | Hixie cpearce roc can you take a look at something I drafted earlier today? |
| 23:32 | <roc> | I see what you're saying, but I think this is something that is easily detected and fixed by the author |
| 23:33 | <rwaldron> | https://gist.github.com/1031549 |
| 23:33 | <roc> | they may not understand why they have to fix it, but that's less important |
| 23:33 | <rwaldron> | It describes a unique behaviour that has value |
| 23:33 | <roc> | I don't think it's worth making the event less useful just to prevent this |
| 23:33 | <rwaldron> | fair enough |
| 23:33 | <rwaldron> | but see above |
| 23:33 | <rwaldron> | or... here https://gist.github.com/1031549 |
| 23:34 | <cpearce> | rwaldron: r-. that can be achieved with canplaythrough + your own tracking code. |
| 23:34 | <rwaldron> | yeah, like I said, we're doing that |
| 23:34 | <roc> | having two events, one which fires once and one which fires more than once, is going to add to the confusion |
| 23:34 | <rwaldron> | i can see there is no interest |
| 23:35 | <rwaldron> | i'm going to save this log |
| 23:35 | <kinetik> | an event that fires only once will be incorrect for the reasons i outlined in the w3.org bug. |
| 23:35 | <rwaldron> | thanks for your time gents |
| 23:35 | <roc> | rwaldron: sorry dude |
| 23:36 | <roc> | I'm all for simplifying things for authors |
| 23:36 | <roc> | the problem in this case would be reduced if Chrome fixed their canplaythrough implementation to be non-brain-dead |
| 23:36 | <TabAtkins> | What we need to do is add the ability to listen to an event only once. jQuery has it, and it's nice and simple and useful. |
| 23:36 | <roc> | then authors would get multiple-firing consistently across browsers and wouldn't think it's a Firefox bug |
| 23:36 | <kinetik> | roc: and Opera. |
| 23:37 | <rwaldron> | we actually closed that FF ticket |
| 23:37 | <roc> | TabAtkins: that is a great idea too |
| 23:42 | <hober> | Whatever happend to Mr. Last Week? |
| 23:45 | <roc> | I got bored |
| 23:45 | <roc> | oops |
| 23:46 | <hober> | heh |
| 23:56 | <pererik> | Hixie: if audioTracks and videoTracks were arrays of track objects you could create a new LocalStream from any number of individual tracks, either by making LocalStream contructable or through some factory |
| 23:58 | <pererik> | Hixie: if tracks are also made available on remote streams, it would be possible to combine several incoming audio streams and record a conversation |
| 23:58 | <pererik> | Hixie: or just recording local audio, while sending both audio and video |
