| 07:17 | <virtuelv> | Wow. Microsoft provided the feedback in a readable form |
| 07:23 | <Dashiva> | Lachy: Yes |
| 07:41 | <Hixie> | if anyone wants to reply to microsoft's feedback, i sent a plain text version with the bs cut out |
| 07:45 | <hsivonen> | I think I have a couple of observations, but perhaps I should refrain from poking this stuff |
| 07:45 | <hsivonen> | (nothing that hasn't been said before) |
| 07:56 | <hsivonen> | whoa! UPnP lets device configuration state be changed by GET with a magic header? that's sad. |
| 08:08 | <Hixie> | uPnP is a disaster |
| 08:09 | <othermaciej> | that is scary |
| 08:09 | <othermaciej> | but I guess that means it is unsafe to allow custom headers beyond a specific whitelist for cross-site requests? |
| 08:09 | <Hixie> | i thought we already had established that |
| 08:09 | <othermaciej> | (or that server-side preflight opt-in is required?) |
| 08:09 | <othermaciej> | I'm not 100% sure of the state of things but I had assumed that was so as well |
| 08:09 | <Hixie> | didn't XHR2 have a whitelist for headers already? |
| 08:10 | <Hixie> | i've kinda lost track of the way the spec is |
| 08:10 | <Hixie> | it keeps bouncing back and forth |
| 08:10 | <othermaciej> | there's server out there that will have side effects based on GET with a query string as well |
| 08:10 | <Hixie> | yup |
| 08:10 | <Hixie> | lots, ven |
| 08:10 | <Hixie> | even |
| 08:10 | <hsivonen> | hmm. one of the key arguments against CS-XHR is that clueless PHP programmers will shoot themselves in the foot if you give them enough rope |
| 08:11 | <othermaciej> | I think I might actually prefer to meet with Mozilla+Opera+Google than the full f2f at this point, to make progress on this topic |
| 08:11 | <hsivonen> | but then the key rationale why XDR doesn't suck is that MS advices people not to shoot themselves in the foot |
| 08:11 | <othermaciej> | hsivonen: that argument applies to XDR as well |
| 08:12 | <Hixie> | othermaciej: i'd be happy to attend such a meeting |
| 08:12 | <othermaciej> | well, maybe we can take time out from the f2f for that |
| 08:12 | <Hixie> | wfm |
| 08:13 | <Hixie> | i guess i should book a hotel |
| 08:13 | <othermaciej> | I don't want to devote a huge amount of time to Microsoft's feedback if they are not interested in coming up with an interoperable solution |
| 08:13 | <othermaciej> | yeah I gotta get that set up as well |
| 09:43 | <Lachy> | MikeSmith, can the bugzilla prefs be changed so that public-html-bugzilla doesn't receive mail for simple things like changing the CC field, and only receives mail for substantial changes/comments |
| 09:44 | <MikeSmith> | Lachy: yeah, sure |
| 09:44 | <MikeSmith> | will do it now |
| 09:45 | <MikeSmith> | I had not really intended originally that the list be something that people actually subscribed to |
| 09:45 | <MikeSmith> | I had thought of it as just being a place to have a record |
| 09:45 | <MikeSmith> | But I can see it needs to be more usefl |
| 09:45 | <MikeSmith> | will change the prefs right now |
| 09:46 | <Lachy> | thanks |
| 09:47 | <Lachy> | I assumed lots of people would subscribe to it, since it's easier than manually adding oneself to individual bugs each time |
| 09:47 | <Dashiiva> | Yeah |
| 09:50 | <MikeSmith> | Lachy: btw, based on your heads-up previously, I did get the list perms set such that only the bugzilla mailbot can post to the list |
| 09:50 | <MikeSmith> | so no more spam |
| 09:55 | <Lachy> | ok, thanks |
| 09:57 | <MikeSmith> | Lachy: OK, mail goes out now only for the following cases: |
| 09:57 | <MikeSmith> | The bug is resolved or reopened |
| 09:58 | <MikeSmith> | New comments are added |
| 09:58 | <MikeSmith> | New attachments are added |
| 09:58 | <MikeSmith> | Some attachment data changes |
| 09:58 | <Dashiiva> | The first case includes new bugs? |
| 10:00 | <MikeSmith> | Dashiiva: yeah |
| 10:01 | <MikeSmith> | actually, there's one more case: |
| 10:01 | <MikeSmith> | I'm added to or removed from this capacity |
| 10:01 | <Dashiiva> | So if someone assigns a bug to the email alias user, we'd get email? :) |
| 10:01 | <MikeSmith> | "added" also covering the case where the address is in the default Cc to begin with |
| 10:01 | <MikeSmith> | Dashiiva: yeah |
| 10:02 | <MikeSmith> | I can't see how to prevent that if someone were to want to be a smartass and do it |
| 10:03 | <Dashiiva> | Just keep it in as a way to detect and filter out smartasses |
| 10:18 | <MikeSmith> | Dashiiva: :) |
| 11:06 | <annevk> | Lachy, the toDataURL line you paste should be added, it should not reply any existing line |
| 11:09 | <Philip`> | Does "[Variadic] in any args" mean any number of arguments, including zero? |
| 11:09 | <heycam> | Philip`, yes |
| 11:09 | <Lachy> | annevk, why? WebIDL says [Variadic] is zero or more |
| 11:11 | <annevk> | :/ |
| 11:11 | <annevk> | that's confusing |
| 11:12 | <Lachy> | why is that confusing? Did you expect it to be at least 1? |
| 11:12 | <annevk> | yes |
| 11:13 | <Philip`> | It makes me think of C functions like "int printf(char*, ...)" where the ... can be zero or more arguments |
| 11:14 | <Lachy> | I wonder if there are any cases where a Variadic requires at least 1 argument, and whether it would be worth adding an argument like [Variadic=0] or [Variadic=1]. But I can't think of any such methods right now. |
| 11:14 | <annevk> | well, making it minimal 1 doesn't hurt anyone |
| 11:14 | <annevk> | as the optional case can be explicitly listed due to overloading |
| 11:14 | <heycam> | though you could just do void f(in int requiredArg, [Variadic] in int optionalArgs) |
| 11:15 | <heycam> | an alternative would be to put the [Variadic] on the operation itself, but at least on the argument you get to keep the type there |
| 11:15 | <heycam> | and a hook for language bindings that don't do varargs |
| 11:15 | <heycam> | (where that [Variadic] argument could map to an array or whatever) |
| 11:17 | <annevk> | (it was also not entirely clear to me Variadic would mean 1 or more, I thought it was just about accepting various types) |
| 11:18 | <annevk> | but maybe I should just read the spec first |
| 11:22 | <annevk> | http://www.w3.org/mid/48561744.40604⊙cc -_- |
| 11:23 | <Lachy> | heycam, why did you call it Variadic? Isn't there a more understandable name that could be used? |
| 11:23 | <heycam> | hmm, i thought it was the appropriate word |
| 11:23 | <heycam> | (being what the "var" stands for in "varargs") |
| 11:23 | <heycam> | [MultiArgs!] :) |
| 11:23 | <Lachy> | [Varargs] would make more sense |
| 11:24 | <hsivonen> | I supposed I should trust that vtab is gone for good from the parsing algorithm |
| 11:25 | <heycam> | (http://en.wikipedia.org/wiki/Variadic_function) |
| 11:26 | <Lachy> | yeah, I found that. I'd never heard of the term before. |
| 11:26 | <Lachy> | http://en.wikipedia.org/wiki/Varargs redirects there, so maybe either is acceptable |
| 11:29 | <roc> | Variadic is a word |
| 11:29 | <roc> | Varargs is not |
| 11:30 | <heycam> | although i'd say varargs might be more widely recognised, i'd go for variadic for the reason roc says |
| 11:30 | <Lachy> | according to which dictionary? |
| 11:30 | <roc> | any dictionary that wasn't written by C hackers |
| 11:30 | <Lachy> | neither answers.com or dictionary.com know of it. |
| 11:31 | <annevk> | they're prolly done in C then! |
| 11:31 | <heycam> | most dictionaries don't have technical terms |
| 11:32 | <Lachy> | ok, I believe you now. http://en.wiktionary.org/wiki/-adic |
| 11:32 | <heycam> | is the root of the word the same as that of "monadic", i wonder? |
| 11:32 | <heycam> | ah there you go :) |
| 11:33 | <roc> | btw you need subject lines in all your IRC messages |
| 11:34 | <heycam> | Subject: Re: btw you need subject lines in all your IRC messages\n\nreally? |
| 11:34 | <annevk> | so you'd say "print() is a niladic function"? |
| 11:35 | <heycam> | well i wouldn't, that sounds silly :) |
| 11:35 | <roc> | Mathematicians actually do talk about nullary functions sometimes |
| 11:35 | <annevk> | or maybe "print() is niladic" |
| 11:36 | hsivonen | wonders if the differences in the notion of whitespace between HTML5, XML 1.0 and XML 1.1 have caused any exploitable TOC/TOU security holes |
| 11:37 | <roc> | hmm, following the -ary suffix, Wikipedia suggests 'multary' or 'multiary' |
| 11:37 | <roc> | variary |
| 11:37 | <heycam> | hehe |
| 11:39 | hsivonen | zaps vtab, is going to be unhappy if it makes a comeback |
| 11:41 | <hsivonen> | Hixie: regarding the generic [R]CDATA algorithm: Gecko seems to be happy to execute a script even if an empty script element is appended to the DOM first and then the entire script is appended as one text node child to it |
| 11:41 | <hsivonen> | WebKit, too |
| 11:42 | <hsivonen> | Hixie: so the unusual point of view of the control where the tree builder starts pulling tokens is unnecessary if the tree builder buffers text node contents |
| 11:42 | <hsivonen> | (unless I'm missing something, of course) |
| 11:43 | <hsivonen> | Hixie: (tested with the tree builder executing in Java but driving browser DOMs through the magic of GWT hosted mode) |
| 11:43 | <hsivonen> | (and the tree builder executing as GWT-compiled JS) |
| 12:08 | <hsivonen> | http://lists.w3.org/Archives/Public/www-tag/2008Jun/0078.html |
| 12:09 | <hsivonen> | I wonder what general assumptions about CDNs are wrong |
| 12:32 | <hsivonen> | does anyone happen to remember what the unresponsive script timeout time is in Gecko and WebKit? |
| 12:33 | <hsivonen> | (does Opera have a timeout?) |
| 12:34 | hsivonen | discovers that the timeout has an unobvious pref name |
| 12:35 | <hsivonen> | seems to be 10 seconds in Firefox 3 |
| 12:35 | <Dashiiva> | security through users never guessing the name of the pref |
| 12:38 | <annevk> | I don't think Opera has a timeout, we don't really need it |
| 12:42 | <virtuelv> | I've had scripts running for 30 minutes |
| 12:42 | <virtuelv> | (Don't ask, just don't) |
| 12:44 | <zcorpan> | virtuelv: was it while(1); ? |
| 12:44 | <virtuelv> | zcorpan: no |
| 12:45 | <virtuelv> | generated some huge-ass images with canvas |
| 12:46 | <annevk> | mandelbrot++ |
| 12:49 | <virtuelv> | yes |
| 12:49 | <virtuelv> | and the implementation sucks, I know |
| 12:50 | <virtuelv> | I should probably rewrite it some day |
| 13:01 | <zcorpan> | are my expectations wrong here? http://simon.html5.org/test/html/dom/interfaces/HTMLElement/HTMLMediaElement/const-unsigned-short/002.htm |
| 13:01 | <zcorpan> | i don't quite grasp ecmascript and webidl :( |
| 13:01 | <zcorpan> | or i don't know the details i guess |
| 13:20 | <hsivonen> | what should I do with exceptions thrown out of code run via setTimeout? |
| 13:20 | <Dashiiva> | Wrap the functions used in setTimeout in a try/catch wrapper? |
| 13:21 | <hsivonen> | Dashiiva: and discard the exception in catch? |
| 13:21 | <Dashiiva> | Depends on what you're trying to accomplish. Which I forgot to ask. |
| 13:22 | <hsivonen> | I'm trying not to reinvent clean error handling patterns for long-running JS code that pumps work units using setTimeout |
| 13:23 | <hsivonen> | hmm. I suppose I make an error callback for the app to set |
| 13:23 | <Dashiiva> | Sounds like a plan. |
| 13:30 | <Philip`> | virtuelv: If you're going to rewrite it, you should rewrite it in C ;-) |
| 13:30 | <Philip`> | Make a viewer like Google Maps, where each Mandelbrot tile is computed on the server, and with no limit on how far you can zoom in |
| 13:32 | <Philip`> | zcorpan: Having a function named 'assert' that sets its argument to 42 is confusingly unconventional |
| 13:34 | <zcorpan> | Philip`: yes |
| 13:35 | <zcorpan> | Philip`: i'll change it to test() :) |
| 16:41 | <Philip`> | Wow, web browser interoperability actually sort of works - Microsoft's fancy AJAX newsgroup reader appears to mostly work fine in Opera 9.5, as long as I set "Mask as Internet Explorer" |
| 16:54 | <annevk> | BTW, there was some confusion about Access Control and headers in this channel at some point. There's indeed a whitelist for GET requests. For headers not on the whitelist a preflight request is made. (This is not a concern for other methods where a preflight request is already a requirement.) |
| 17:00 | <annevk> | It seems that it is also part of the MS feedback though the later admit it's protected by a preflight request. I don't think their feedback identifies any new issues. (I've read through it twice so far.) |
| 17:39 | <gsnedders> | huh |
| 17:39 | <gsnedders> | jgraham: is http://james.html5.org/temp/outline/outline.py out of date? |
| 17:39 | <gsnedders> | jgraham: I'm getting different results running it locally :\ |
| 17:40 | <gsnedders> | <h1>Foo</h1><h2>Bar</h2><h2>Lol</h2> gives what I'd expect at <http://james.html5.org/outliner.html>;, but running it locally agrees with my impl, which makes the second <h2> at the same level as the <h1> |
| 18:54 | <Dashiva> | Anyone involved with bindings4dom here? |
| 18:56 | <Dashiva> | I'm reading 4.4.2. Host object [[Put]] method. There seems to be no way to set a property that doesn't already exist. |
| 19:02 | <annevk> | heycam is the editor |
| 23:03 | <heycam> | Dashiva, if that's the case that's a bug :) |
| 23:03 | <heycam> | i'll have a look once i'm at uni |
| 23:12 | <jgraham__> | gsnedders: I copied the latest version of outline.py to james.html5.org/temp/outline/outline.py |
| 23:36 | <Dashiva> | Just to see if I got the terminology right... HTMLDivElement is an 'interface object', HTMLDivElement.prototype (also somediv.[[prototype]]) is an 'interface prototype object' and somediv is a 'host object implementing an interface'. |
| 23:36 | <heycam> | Dashiva, i see the bug (the "go to step 23") |
| 23:36 | <heycam> | (in web idl's [[Put]]) |
| 23:37 | <heycam> | Dashiva, yeah that's the terminology i used |
| 23:37 | <Dashiva> | heycam: That's half of it |
| 23:37 | <Dashiva> | The other half is step 14 |
| 23:38 | <Dashiva> | If the property doesn't exist, it continues to 15 and from there you can't get anywhere beyond 19 |
| 23:38 | <heycam> | hmm |
| 23:39 | <heycam> | ok i'll take a look at it |
| 23:39 | <Dashiva> | Actually, probably step 15 is the problem |
| 23:39 | <Dashiva> | Since it has to check PutForwards before going on to create |
| 23:39 | <Dashiva> | So instead of throwing an exception, it should jump to 24 |
| 23:40 | <Dashiva> | *19 |
| 23:40 | <Dashiva> | You know what, it's too late for this. Don't listen to me. Zzz :) |
| 23:41 | <heycam> | yeah that step 15 is assuming the property already exists |
| 23:41 | <heycam> | which it shouldn't |