| 00:03 | <Philip`> | Has anybody already mentioned that <canvas> ImageData won't work with high-end display systems that use 12-bit colour components, or floating point components? |
| 00:06 | <SadEagle> | I guess it should permit better precision in the same range? |
| 00:07 | <Philip`> | That wouldn't help the floating-point HDR case, though maybe that case is a bit too theoretical |
| 00:08 | <SadEagle> | I suppose the clamping is the right behavior for a non-HDR implementation on HDR data, or am I mistaken? |
| 00:09 | <Philip`> | Depends on whether you consider data loss to be "right behavior" :-) |
| 00:11 | <annevk> | css has the same issue |
| 00:11 | <SadEagle> | well, premultiplied alpha is data loss, too .... |
| 00:11 | <annevk> | although CSS has the weird thing that it allows percentages too which allow for a far larger range than the integer alternative |
| 00:12 | <Philip`> | When we all have cyborg eyes that can detect four distinct colours, all this three-component RGB legacy will be a real pain |
| 00:12 | <SadEagle> | It definitely seems like something that should be permitted. So if an implementation supports 9 bits,and is given 179.5, it can represent it well. |
| 00:57 | <roc> | othermaciej: what was the soft-hyphen problem with following the spec? |
| 00:58 | <roc> | if there's some site depending on us not following the spec, I'll make sure we break them in Firefox 3 :-) |
| 00:58 | <othermaciej> | roc: the ECMAScript 3 spec says that soft hyphens in JS source should be skipped, but Mozilla doesn't do that (in fact the Mozilla JS regression tests had a test in the ecma_3 directory that tests for not doing it) |
| 00:58 | <othermaciej> | roc: I don't think dropping soft hyphens is a useful behavior anyway |
| 00:58 | <othermaciej> | it should just be dropped in ES4 |
| 00:58 | <roc> | ah ok, JS, sorry, can't help you there |
| 00:59 | <othermaciej> | roc: I'm not sure breaking sites to follow the spec is always be best approach in any case |
| 00:59 | <othermaciej> | depends on the bug and the site |
| 00:59 | <roc> | yep |
| 00:59 | <Dashiva> | Philip`: Isn't part of the motivation for imagedata being able to fit it into bytes for an bytearray? |
| 01:00 | <othermaciej> | in that way I guess it's healthier to be doing testing and development on specs where a new version is under active development, so things can be fixed in the spec too |
| 01:00 | <roc> | specs need maintenance too |
| 01:02 | <Philip`> | We need an acid test for specs, to detect when the specs are broken |
| 01:02 | <othermaciej> | yes and it's hard to develop against an unmaintained spec |
| 01:02 | <Dashiva> | Philip`: First we need a spec spec to say what's a conforming spec |
| 01:02 | <othermaciej> | I'd say DOM Core is probably the most unmaintained now |
| 01:02 | <othermaciej> | although for ES4 it's very hard right now to tell how much it is fixing ES3 spec bugs |
| 02:05 | <Lachy> | interesting. IE6 supports security=restricted attribute for iframes http://blogs.msdn.com/ie/archive/2008/01/18/using-frames-more-securely.aspx |
| 02:06 | <othermaciej> | I'm curious how much detail is available about what they restrict |
| 02:06 | <othermaciej> | it might be a useful thing for ad frames or the like |
| 02:07 | <othermaciej> | or if you were willing to put user-generated content in an iframe you could use it for XSS robustness |
| 02:07 | <webben> | othermaciej: It says: "Frames running in the Restricted Sites zone cannot run script" so I'm guessing: not very useful. |
| 02:08 | <othermaciej> | webben: I guess it depends on how much of what your ad server does is client-side |
| 02:08 | <othermaciej> | I guess even a regular iframe off of a different domain is a useful security restriction |
| 02:08 | <webben> | Sure. But the typical graphical ad seems very JS-dependent. |
| 02:08 | <othermaciej> | they're either animated gifs or Flash |
| 02:08 | <othermaciej> | few of them use JS as part of the ad per se |
| 02:09 | <SadEagle> | a huge number use JS to boostrap it |
| 02:09 | <webben> | indeed |
| 02:09 | <webben> | pretty much all of them on the network I work on --- although most have a noscript fallback or whatever. |
| 02:10 | <webben> | that's not to say they genuinely require JS for what they do; but changing what ad authors do isn't necessarily easy |
| 02:10 | <webben> | (there's also the issue of JS used for click tracking etc) |
| 02:12 | <SadEagle> | Hmm, does window.location on a frame actually navigate the entire window, though? |
| 02:13 | <othermaciej> | no |
| 02:14 | <othermaciej> | it navigates the frame |
| 02:15 | <SadEagle> | I thought so :-). yeah, so the example in the blog is wrong. |
| 02:49 | <weinig_> | SadEagle: it could do top.location though |
| 02:49 | <SadEagle> | weinig_: hmm, right, the XSS models permits setting of location href :( |
| 09:35 | <jruderman> | Hixie: safari messes up http://www.libpr0n.com/ :( |
| 09:35 | <jruderman> | Hixie: i think acid3 should test whatever it is that makes safari mess it up |
| 09:36 | <jruderman> | clearly something related to the "@ debug" thing |
| 09:39 | <othermaciej> | jruderman: @ debug; * { border: 1px solid white ! important; } /* see below */ is matching |
| 09:39 | <othermaciej> | I'm not sure whether it should or not |
| 09:40 | <othermaciej> | as I do not know what is expected to terminate a bad @ rule |
| 09:40 | <othermaciej> | I am not sure why it uses such a weird-ass technique to comment that line |
| 09:40 | <jruderman> | "see below" has a long explanation |
| 09:41 | <jruderman> | it's there to give the site author an easy way to toggle that rule on and off (by switching between "@debug" and "@ debug") |
| 09:41 | <othermaciej> | crazy stuff |
| 09:42 | <jruderman> | yes, crazy stuff |
| 09:42 | <othermaciej> | I am too lazy to check the spec references |
| 10:09 | <annevk> | at-rules are ; or { }-block terminated |
| 10:09 | <annevk> | so Safari is buggy |
| 10:11 | <annevk> | seems that Opera pre-9.5 got this wrong too |
| 10:11 | <annevk> | maybe pre-9.3 hmm |
| 10:12 | <Ketsuban> | I wasn't aware @debug was a valid at-rule. |
| 10:12 | <annevk> | it's not |
| 10:14 | <Ketsuban> | So changing @ debug to @debug ought to have no effect. |
| 10:14 | <annevk> | it should have effect |
| 10:15 | <Philip`> | Changing from @ debug to @debug changes it from invalid selector to invalid at-rule, which is different |
| 10:15 | <annevk> | there are generic parsing rules for at-rules |
| 10:15 | annevk | initially thought the trick was @debug versus @debug; |
| 10:16 | <Ketsuban> | "User agents must ignore an invalid at-keyword together with everything following it, up to and including the next semicolon (;) or block ({...}), whichever comes first." |
| 10:20 | <Ketsuban> | Either way this is a really stupid "trick" - if you want to "quickly switch on/off a style" comment/uncomment it. |
| 10:22 | <annevk> | it's just a hack |
| 10:23 | <annevk> | and probably not widely known |
| 10:25 | <Philip`> | It's probably intended to break as many non-Mozilla browsers as possible |
| 10:27 | <annevk> | yeah |
| 10:31 | <jruderman> | hehe |
| 10:31 | <jruderman> | who owns libpr0n.com? hixie? |
| 10:34 | <jruderman> | http://libpr0n.com/links.html tries to link a porn site, but apparently its registration expired and now it displays a non-porn squatter page. what irony ;) |
| 10:43 | <krijnh> | Wow, I didn't disconnect :) |
| 11:00 | <othermaciej> | the actual CSS file contains "@ debug;" |
| 11:01 | <othermaciej> | the trick is that this is the start of a bad selector, not a bad @rule |
| 13:55 | <gsnedders> | a feed served as text/rdf. fun. |
| 14:02 | Philip` | wonders what makes it fun |
| 14:03 | <gsnedders> | Philip`: doing magic tricks working out what content is :) |
| 14:04 | <gsnedders> | (also note text/rdf is an unregistered MIME type, RDF should be application/rdf+xml) |
| 14:06 | <Philip`> | Having seen a "Content-Type: ��~/html;charset=UTF-8", I won't consider any content-type fun if it merely has the wrong letters in it |
| 14:07 | <Philip`> | Also, argh, those characters break my IRC display |
| 14:08 | <gsnedders> | :P |
| 14:28 | <Ketsuban> | Philip`: Can you give Unicode references for those? They got replaced with U+FFFD REPLACEMENT CHARACTER. |
| 14:31 | <Philip`> | Ketsuban: I can't, since I only see U+FFFD and I can't just redownload the headers since they're random |
| 14:31 | <Philip`> | but see HEAD http://www.louvre.fr for a live demonstration :-) |
| 14:32 | <Ketsuban> | Looks normal to me. |
| 14:33 | <gsnedders> | looks fine to me too |
| 14:33 | <Philip`> | Using HEAD, not GET? |
| 14:34 | <annevk> | doesn't look normal to me |
| 14:34 | <Ketsuban> | If I knew what "using HEAD, not GET" meant I might be able to tell. :P |
| 14:34 | <gsnedders> | using HEAD. |
| 14:35 | <zcorpan> | http://www.rexswain.com/httpview.html says Content-Type:·text/html;·charset=iso-8859-1(CR)(LF) |
| 14:35 | <Philip`> | Oh, maybe my HEAD is following redirections |
| 14:35 | <Philip`> | Try it with http://www.louvre.fr/llv/commun/home.jsp instead |
| 14:35 | <zcorpan> | i tried with both follow and not follow redirects with rexswain |
| 14:36 | <Ketsuban> | Content-Type:·(B0)(HT)B(03)/html;charset=UTF-8(CR)(LF) |
| 14:36 | <annevk> | does that use HEAD requests though? |
| 14:36 | <Philip`> | Content-Type:·8(CC)D(06)/html;charset=UTF-8(CR)(LF) |
| 14:36 | <zcorpan> | ah yep |
| 14:36 | <Philip`> | Ketsuban: "(HT)" ? |
| 14:36 | <annevk> | I get a different response each time |
| 14:36 | <Philip`> | Oh, horizontal tab, not weird hex |
| 14:37 | <zcorpan> | ah, i didn't look at "Location 2" ... |
| 14:37 | <annevk> | �`D/html;charset=UTF-8 and P'/html;charset=UTF-8 ... |
| 14:37 | <Philip`> | annevk: Now you've made my IRC display blink :-( |
| 14:38 | <zcorpan> | ������������������������������ |
| 14:45 | <didymos> | Philip`, Yeah, it's quite annoying |
| 15:12 | <harri> | same here. funny that it's still possible in 2008. |
| 17:22 | <gsnedders> | annevk: re: dragging it up again: it never went away |
| 20:41 | <zcorpan> | http://sitegod.blogspot.com/2008/01/alright-i-come-clean-i-hate-html-5.html |
| 20:50 | <roc> | lame |
| 20:50 | <roc> | It's good to hear we control the W3C though |
| 20:51 | <SadEagle> | indeed lame. I don't agree with some things in html5, but I can't argue with the preference of getting useful features in there, instead of taking part in self-indulgent dances going nowhere. |
| 20:53 | <jasonw22> | it's too bad folks don't limit themselves to the technical details in this discussion, and resort to hysterical vitriol |
| 20:55 | <gsnedders> | And get the technical details wrong |
| 20:56 | <SadEagle> | uhm, the guy is 16 :-) |
| 20:56 | <gsnedders> | SadEagle: uhm, I'm 15 |
| 20:56 | <jasonw22> | ah, 'nuf said |
| 20:57 | <gsnedders> | So I guess I'm as dumb as him, just because of my age. |
| 20:57 | <SadEagle> | gsnedders: that's not what I meant. I mean that some people of your age are not quite mature yet. |
| 20:58 | <Philip`> | Some people of any age aren't :-) |
| 20:59 | <gsnedders> | … which is the entire issue of basing it on age |
| 20:59 | <SadEagle> | well, but they should be held to a somewhat higher standard :-) |
| 20:59 | SadEagle | ponders shutting his mouth before he digs an even bigger hole. |
| 21:02 | <gsnedders> | on the subject of digging your way out of holes, I ought to write a post about local branch of Tesco |
| 21:06 | <gsnedders> | <http://www.fifetoday.co.uk/news/Madras-pupils-voice-anger-over.3369908.jp>; — very shortened copy of the article in newspaper |
| 21:06 | <Philip`> | There was almost a new Tesco store a couple of years ago around where I lived, built on top of a railway tunnel, but it got delayed a bit because the tunnel collapsed and they haven't restarted building yet |
| 21:07 | <gsnedders> | That was silly, to say the least. |
| 21:08 | <Philip`> | (Fortunately there weren't any trains there at the moment it collapsed, but it did cause quite a lot of disruption) |
| 21:09 | <gsnedders> | the whole situation here was far more bizarre than that article leads to suggest |
| 23:02 | <gsnedders> | http://pastebin.ca/867866 should fix your parse error Philip` as well as one or two other issues |
| 23:05 | <gsnedders> | (again, and further comments are welcome) |
| 23:05 | <gsnedders> | (and, before it's asked: the discussion about the issue never ended sadly, this won't make anything worse happen :P) |