| 00:33 | <Hixie> | the arguments saying that latex is more useful than anything else are realy arguing that it isn't alternative text but is better than the image itself |
| 00:41 | <Lachy> | Hixie, I don't think so. |
| 00:43 | <jcranmer> | I think they're working along the lines that "anyone who knows anything more complex than exponentiation knows (La)TeX", something which seems dubious to me |
| 00:44 | <jcranmer> | ideally, it could be read aloud, but conveying |
| 00:46 | <jcranmer> | \lim_{n \to \inf} \left[ \sum_{k=0}^\inf \frac{n^k}{\left(k+1\right)^{k+1}} \right]^{\frac{1}{n}} |
| 00:46 | <jcranmer> | is probably not fun |
| 00:47 | <jcranmer> | "The limit as n goes to infinity of the sum from k equals 0 to infinity of n to the k over k plus 1 to the k + 1 power all to the 1 over n power" |
| 00:54 | <Lachy> | both of those are complicated. That latter is even a little ambiguous in some parts |
| 00:56 | <Lachy> | e.g. does "k plus 1 to the k + 1 power" mean: k + 1^(k+1) or (k + 1)^(k+1) |
| 00:57 | <jcranmer> | that's the problem with spoken stuff: it's hard to get parenthetics rights |
| 00:57 | <Hixie> | if i was reading those, the former would be "k plus one to the k plus oneth power", the latter would be "k plus one, all raised to the k plus oneth power". |
| 00:58 | <Hixie> | jcranmer: that's a problem with english, not just spoken word |
| 00:58 | <Lachy> | so we should all start speaking in TeX instead :-) |
| 00:59 | <jgraham> | FWIW T.V. Ramen's LaTeX->speech translator is supposed to set the inonation and pauses to effectively convey the meaning. You can't do that with pure english |
| 00:59 | <Hixie> | those could also be read as "k plus one to the open bracket k plus one close bracket" and "open breack k plus one close breacket to the open breack k plus one close bracket" |
| 00:59 | <Hixie> | wow i can't spell bracket |
| 00:59 | <Hixie> | wtf |
| 00:59 | <jgraham> | s/inonation/intonation/ |
| 01:00 | <jgraham> | s/pure english/pure english conveyed as ascii text/ |
| 01:00 | <Hixie> | you know, if the fallback is going to be some format language, maybe <object data=eq.png><object data=eq.latex>...</object></object> is a better solution |
| 01:00 | <Lachy> | you can't spell "break" either (unless you meant bracket there too) |
| 01:01 | <Hixie> | i meant bracket each time |
| 01:01 | <Lachy> | ok |
| 01:01 | <Hixie> | i really can't type it apparently :-) |
| 01:01 | <jcranmer> | :-) |
| 01:01 | <jcranmer> | +1 on the nesting object idea |
| 01:02 | <jgraham> | I don't see why this is a big problem. Latex is an ascii (or at least text-based) equivalent of a formula. When that formula is contained in an image the LaTeX is therefore an equivalent of that image |
| 01:02 | <Lachy> | why don't we just add a way to use latex natively in HTML? |
| 01:03 | <jgraham> | Lachy: Because that would be a big mess |
| 01:03 | <jcranmer> | I thought that was called MathML? ;-) |
| 01:03 | jcranmer | knows the two aren't really alike |
| 01:03 | <Hixie> | hat jcranmer said |
| 01:03 | <jgraham> | However I already suggested that we add a TeX-like syntax that maps to MathML |
| 01:03 | <Hixie> | mathml is our solution for math |
| 01:03 | <Lachy> | most mathematicians seem to use latex though anyway |
| 01:03 | <Hixie> | we already have a syntax |
| 01:03 | <jgraham> | Hixie rejected it |
| 01:04 | <jgraham> | Lachy: This is because TeX doesn't suck and MathML does |
| 01:04 | <jcranmer> | Lachy: there are numerous LaTeX->MathML converters |
| 01:04 | <jgraham> | (well and also because TeX has meaningful deployment) |
| 01:04 | <jcranmer> | and numerous MathML->TeX converters |
| 01:04 | <Lachy> | so why was mathml created in the first place, instead of just defining how to add latex to HTML and other XML languages? |
| 01:05 | <jgraham> | jcranmer: There are numerous (*subset of* LaTeX)->(MathML) convertors |
| 01:05 | <jcranmer> | I presume if you stick a lot of \kern + co in there, it might not translate |
| 01:06 | <jgraham> | Lachy: Well you should ask David Carslile (sp?) but I think a) people wanted to get rid of some TeX warts b) people thought they needed semantics c) XML is cool |
| 01:07 | <jgraham> | jcranmer: Or anything that relies on something you \def yourself |
| 01:07 | <jcranmer> | jgraham: a smart converter should be able to de-macro \def's |
| 01:08 | <jgraham> | jcranmer: By "smart convertor" you mean one that actually runs the TeX right? |
| 01:09 | <jcranmer> | that is ideal |
| 01:10 | <jgraham> | I think you would still have some difficulty because to do a good job at creating MathML you need to retain some element of what the user typed |
| 01:10 | <jgraham> | (to have a shot at the semantics) |
| 01:12 | <jgraham> | even then getting things like operator vs identifier right is hard |
| 01:13 | <jgraham> | it seems much more sensible to define a reduced subset of LaTeX that maps well onto MathML in the same way PDF is a subset of postscript |
| 01:22 | <jcranmer> | quick, OT question: if you were to create a utility that reposts spamified messages from a newsgroup of certain users, would you expect that the tools you use are able to properly support =?utf-8?Q?...?= in MIME headers? |
| 10:08 | <hsivonen> | hmm. people want a category for markp's posts as if there were too many others posts to clutter their feed readers |
| 10:29 | <Lachy> | hsivonen, yeah, I thought it was a bit odd too. But it's not hard to do |
| 12:19 | <Lachy> | Hixie, I contest your claim that "the main characters [or Stargate] aren't immortal". IIRC, Daniel has died 4 times, Jack, Carter and Teal'c have died once, and yet each time they have been resurrected. And this doesn't include alternate realities and timelines. :-) |
| 12:20 | <Lachy> | s/[or/[of/ |
| 14:14 | <jacobolus> | hsivonen, Lachy: you could alternately make a thing at Yahoo Pipes that filters out posts other than MarkP's |
| 14:30 | <jacobolus> | e.g. http://pipes.yahoo.com/pipes/pipe.run?_id=47fd026d4015483a48d8cd5656c0636e&_render=rss |
| 14:37 | <jacobolus> | better: http://pipes.yahoo.com/jrus/this_week_in_html5?_render=rss |
| 16:32 | hsivonen | guesses that someone will implement a pure-JS DOM for Web Worker XHR |
| 16:56 | <jgraham> | hsivonen: On the podcast thing he id recently John Reisig was talking about something like having access to insulated DOM-fragments in workers and a browser-level protocol for passing those fragments through channels to the main thread |
| 16:56 | <jgraham> | Sorry too many i's in Resig there... |
| 16:57 | <takkaria> | jgraham: there's some code to bind hubbub to libxml2 now; I don't know quite how much use that will be to you, but thought you would be interested |
| 17:00 | <Lachy> | jacobolus, what rules did you use to make that? |
| 17:01 | <Lachy> | jacobolus, btw, I suspect making a category would be easier. I just need to know what to call it |
| 17:01 | <jgraham> | takkaria: That sounds awesome. I will have a look when I have a spare moment (which may not be for some time...) |
| 17:02 | <takkaria> | jgraham: when you do have a spare moment, give me a yell. I'll try and sort it out so the code is somewhere public before then. :) |
| 17:04 | <Lachy> | jacobolus, I found the yahoo pipe source. Filtering on the name is only reliable if markp is the only one who ever writes them |
| 18:40 | <jacobolus> | Lachy: it could be filtered on anything else, if you like, even on a regexp or something |
| 18:40 | <jacobolus> | Lachy: but you're right that a category would be easiest—I just don't have control over that :) |
| 18:54 | <Lachy> | jacobolus, I do. I just need to know what to call the category |
| 18:55 | <Lachy> | any suggestions? maybe weekly-review |
| 18:55 | <jacobolus> | recent-highlights? |
| 18:55 | <jacobolus> | weekly-review sounds fine |
| 18:55 | <jacobolus> | if you can actually keep it weekly ;) |
| 20:20 | <hsivonen> | zcorpan: Mail::Mailer brokenness fixed. thanks. |
| 20:39 | <hsivonen> | oops Mail::Mailer still broken. sigh. |
| 20:48 | <hsivonen> | where might CPAN break stuff? I remove perl stuff from under /usr/local, but Mail::Mailer still fails due to a tainted variable in the -T mode |
| 20:48 | <hsivonen> | and the Mail::Mailer should be provided by Ubuntu |
| 20:48 | <hsivonen> | hmm |
| 21:08 | <Hixie> | Lachy: ford. |
| 21:09 | <Hixie> | Lachy: also, they can't very well be immortal if they died! |
| 21:09 | hsivonen | regrets invoking CPAN |
| 21:10 | <hsivonen> | I have no idea how to debug why a variable is tainted in Bugzilla after CPAN has done something |
| 21:12 | <Hixie> | wow, smylers' e-mail was pretty much spot on as far as my current thinking is concerned :-) |
| 21:14 | <Lachy> | Hixie, what do you mean by ford? |
| 21:14 | <Hixie> | he died. |
| 21:15 | <Hixie> | well, one assumes he died, anyway. |
| 21:15 | <Lachy> | oh |
| 21:15 | <Hixie> | one minute he was on a ship, then next minute the ship was in a bazillion pieces, and we never heard from him again. |
| 21:36 | <annevk> | Lachy, I think the blog should have tags |
| 21:36 | <annevk> | Lachy, and we should get rid of the categories |
| 21:41 | <Lachy> | annevk, tags are already supported. We just need to expose them in the template |
| 21:42 | <annevk> | with "the blog should have tags" I mean that existing entries should have tags added and that they (obviously) should be exposed through the Web interface :) |
| 21:43 | <annevk> | then we can just make up some weird tag for markp's stuff and people can rejoice |
| 21:43 | <annevk> | (though given the low amount of posts I wonder why they just want to follow those, especially given only one appeared so far) |
| 21:44 | <Hixie> | does outerHTML have any stupid quirks anyone knows about? |
| 21:44 | <annevk> | in two out of three browsers supporting it stuff is in uppercase? |
| 21:45 | <annevk> | (not sure about WebKit) |
| 21:45 | <Hixie> | i meant for setting more than getting |
| 21:47 | <Hixie> | though thanks for the getting info |
| 21:54 | hsivonen | feels powerless in face of perl -T |
| 21:55 | <Hixie> | do you need to untaint something forcibly? |
| 21:55 | <Hixie> | or are you trying to work out where the taint started, so you can stop that? |
| 21:55 | <hsivonen> | Hixie: the latter |
| 21:56 | <hsivonen> | specifically, if my bugzilla installation worked before touching cpan, why does it stay broken after I delete all perl stuff under /usr/local ? |
| 21:57 | <hsivonen> | now every time Bugzilla tries to send email, the message is considered tainted when it's passed to exec for sendmail |
| 21:57 | <Hixie> | can you work out which part of the message gets considered tainted? |
| 21:57 | <Hixie> | then again, that sounds about right, i mean, that data is coming from the database right? |
| 21:57 | <Hixie> | so it should be tainted |
| 21:58 | <hsivonen> | Hixie: I don't know how to work it out with reasonable effort. |
| 21:58 | <hsivonen> | Hixie: sure, but presumably before I touched cpan, something untainted the message properly |
| 22:00 | <Hixie> | what version are you on? |
| 22:00 | <hsivonen> | 2.22.1-debian2, so I can't get help from bugzilla hackers |
| 22:01 | <Hixie> | and what's the error message exactly? |
| 22:01 | <hsivonen> | undef error - Insecure dependency in exec while running with -T switch at /usr/share/perl5/Mail/Mailer/sendmail.pm line 22. |
| 22:01 | <hsivonen> | line 22 is exec( $exe, '-t', @$args ); |
| 22:02 | <hsivonen> | and I already checked that $exe is not the problem |
| 22:02 | <hsivonen> | so something in $args is tainted |
| 22:03 | <Hixie> | are you sure $ENV{PATH} is cleared? |
| 22:03 | <hsivonen> | Hixie: I'm not. How do I find out? |
| 22:03 | <Hixie> | stick $ENV{PATH} = ''; somewhere before the exec call |
| 22:06 | <hsivonen> | Hixie: that doesn't solve the problem |
| 22:06 | <Hixie> | ok just before the exec() call try this: |
| 22:07 | <Hixie> | use Scalar::Util; warn("exe:" . tainted($exe) . "; args: " . tainted("@$args")); |
| 22:07 | <Hixie> | see if either of those returns true |
| 22:08 | <hsivonen> | where do the warnings end up? |
| 22:08 | <Hixie> | (if you're on an old perl (pre 5.8) you can define sub tainted() { return ! eval { eval("#" . substr(join("", @_), 0, 0)); 1 } } ) |
| 22:08 | <Hixie> | stderr |
| 22:08 | <Hixie> | apache error log probably |
| 22:08 | <Hixie> | failing that, print it to some known location |
| 22:09 | <Hixie> | it should go to the same place as you see the insecure dependency message |
| 22:11 | <hsivonen> | undef error - Undefined subroutine &Mail::Mailer::sendmail::tainted called at /usr/share/perl5/Mail/Mailer/sendmail.pm line 23. |
| 22:11 | <hsivonen> | and I'm not on old perl |
| 22:11 | <Hixie> | sorry make that use Scalar::Util qw(tainted); |
| 22:12 | <hsivonen> | process_bug.cgi: exe:0; args: 1 at /usr/share/perl5/Mail/Mailer/sendmail.pm line 22. |
| 22:13 | <Hixie> | ok, so $args is being tainted somehow |
| 22:14 | <Hixie> | use Scalar::Util qw(tainted); foreach (@$args) { warn tainted($_) } will tell you which item in $args |
| 22:15 | <hsivonen> | 0 and 1 |
| 22:15 | <hsivonen> | wouldn't that be to and from? |
| 22:15 | <Hixie> | no idea |
| 22:15 | <Hixie> | but you can probably look in the code to see |
| 22:16 | <hsivonen> | Hixie: thanks |
| 22:16 | <Hixie> | np |
| 22:19 | <zcorpan> | is whatever feature we come up with to distinguish images-with-proper-replacement to images without going to be used correctly enough that UAs will do something special with it? |
| 22:19 | <zcorpan> | why can't we just allow alt="photo" and be done with it? |
| 22:20 | <zcorpan> | (though when i think about it, opera does something special with lack of alt -- it shows "Image") |
| 22:23 | <Hixie> | i have no diea |
| 22:23 | <Hixie> | idea |
| 22:23 | <Hixie> | it's somewhat up to the accessibility "experts" at this point |
| 22:23 | <Hixie> | if it isn't implemented, then it'll get dropped at CR |
| 22:23 | <Hixie> | and we'll be back to optional alt |
| 22:24 | <zcorpan> | the old spec text didn't allow the kind of image to be specified as alt |
| 22:25 | <Hixie> | indeed, that would be quite bad for accessibility |
| 22:26 | <zcorpan> | why? |
| 22:26 | <Hixie> | you couldn't distinguish images that were photos from images that represented icons of photos |
| 22:26 | <Hixie> | makes for a pretty confusing ui |
| 22:27 | <zcorpan> | i think you could figure that out based on context |
| 22:27 | <Hixie> | *shrug* |
| 22:27 | <Hixie> | can anyone think of any way to get a hold of any of the nodes that are created in foo.outerHTML = '...' if 'foo' has no parent node? |
| 22:28 | <zcorpan> | implicit document fragment node? not sure i understand the question |
| 22:30 | <Hixie> | if i do: <script> var d = document.createElement('div'); d.outerHTML = '<form id=a>'; </script> how can i get hold of the form element? |
| 22:30 | <Hixie> | there's no way, right? |
| 22:32 | annevk | can't say no for certain |
| 22:36 | <zcorpan> | seems so... so outerHTML is basically a no-op in that case |
| 22:36 | <Hixie> | yeah |
| 22:36 | <Hixie> | good |
| 22:37 | <Hixie> | that way i don't have to worry about what its parentNode is to get a context node... |
| 22:38 | <annevk> | in opera if you do d.outerHTML afterwards it says "<form id=a>" |
| 22:38 | annevk | has no idea what that means |
| 22:38 | <Hixie> | lol |
| 22:38 | <Hixie> | that's funny |
| 22:39 | <annevk> | (it also says d is an HTMLDivElement object) |
| 22:39 | <Hixie> | maybe it's setting the actual JS property and not the DOM attribute |
| 22:39 | <Hixie> | try setting it to something that's not HTML syntax |
| 22:39 | <Hixie> | and see if it fixes it up |
| 22:39 | <Hixie> | or if it just allowed anything |
| 22:39 | <Hixie> | e.g. <fOrM |
| 22:39 | <Hixie> | does it become <form> ? |
| 22:40 | <annevk> | no |
| 22:40 | <annevk> | "x" stays "x" |
| 22:40 | <Hixie> | then it's not really outerHTML |
| 22:40 | <annevk> | "<x" stays "<x" |
| 22:41 | <annevk> | setting outerHTML in other cases does work |
| 22:41 | <annevk> | confusing |
| 22:42 | <zcorpan> | in ie it's <DIV></DIV> |
| 22:43 | <annevk> | parentNode == null probably makes browsers invoke special codepaths |
| 22:43 | <annevk> | so outerHTML is going to make it in after all? |
| 22:43 | <Hixie> | everyone except FF implements it |
| 22:43 | <Hixie> | might as well |
| 22:44 | <Hixie> | i'm not doing it for XML though |
| 22:44 | <annevk> | you know you use such arguments the other way around as well, right? |
| 22:44 | <annevk> | :) |
| 22:44 | <Hixie> | depends how hard it is to spec :-P |
| 22:45 | <Hixie> | bigger problem is what to do when the parent is a Document object |
| 22:45 | <zcorpan> | webkit throws when there's no parent it seems |
| 22:46 | <annevk> | per some old testcase here http://tinymce.moxiecode.com/opera/documentelement_outerhtml.htm it should do something |
| 22:47 | <annevk> | maybe the same as document.innerHTML |
| 22:47 | <Hixie> | no, that would blow away comments and doctypes |
| 22:47 | <zcorpan> | webkit and ie throw when parent is document |
| 22:47 | <Hixie> | sounds good to me! |
| 22:49 | <annevk> | zcorpan, they don't show anything in the iframe for that test? |
| 22:49 | <Hixie> | btw html5 has a dependency on xhr now :-) |
| 22:50 | <Hixie> | (well, web workers, technically) |
| 22:50 | annevk | finds http://en.wikipedia.org/wiki/Ian_Hickson |
| 22:50 | <Hixie> | haha sweet, i actually have a wikipedia page now |
| 22:51 | <Hixie> | awww, no controversy in the talk page |
| 22:51 | <Hixie> | how sad |
| 22:51 | <zcorpan> | annevk: right |
| 22:52 | <Hixie> | i guess the right exception is INVALID_ACCESS_ERR here |
| 22:52 | <Hixie> | hough NO_MODIFICATION_ALLOWED_ERR, NOT_SUPPORTED_ERR and HIERARCHY_REQUEST_ERR are all tempting |
| 22:52 | <zcorpan> | log: Error: NO_MODIFICATION_ALLOWED_ERR: DOM Exception 7 |
| 22:52 | <zcorpan> | webkit |
| 22:52 | <Hixie> | sounds good to me! |
| 22:53 | <Hixie> | i never get to throw that one |
| 22:53 | <zcorpan> | hah |
| 22:53 | <Hixie> | does IE throw for no prent node too? |
| 22:53 | <Hixie> | parent |
| 22:53 | <zcorpan> | no |
| 22:53 | <Hixie> | ok |
| 22:54 | <zcorpan> | neither does opera |
| 22:55 | <annevk> | is HIERARCHY not better? though maybe I don't care |
| 22:56 | <zcorpan> | fwiw, ie throws if you try to set outerHTML on head or body too |
| 22:56 | <zcorpan> | or innerHTML on the root |
| 22:56 | <zcorpan> | but we don't align with ie very much for innerHTML |
| 22:58 | zcorpan | tests documentfragment |
| 23:01 | <zcorpan> | opera does the same thing as it does for document |
| 23:01 | <zcorpan> | webkit throws |
| 23:02 | <zcorpan> | ie does the same thing as it does for a normal element parent |
| 23:02 | zcorpan | likes the ie approach |
| 23:02 | <annevk> | I hope I can make fixes and such to XHR / AC / Media Queries tomorrow |
| 23:03 | <Hixie> | don't break my dependency! :-P |
| 23:03 | <annevk> | I'll be off for the rest of the week and next week is the CSS WG meeting... |
| 23:03 | <annevk> | summer is slow anyways I suppose |
| 23:03 | <hsivonen> | zcorpan: please let me know if you see the taint mode message with bugzilla.validator.nu. |
| 23:03 | <Hixie> | oh, let me know how the csswg meeting goes |
| 23:03 | annevk | looks for the depedency |
| 23:03 | <Hixie> | i use NETWORK_ERR |
| 23:03 | <jacobolus> | what does the CSS WG plan to do at their meeting? |
| 23:03 | <jacobolus> | maybe figure out hypenation? ;) |
| 23:04 | <jacobolus> | erm, hyphenation |
| 23:04 | <hsivonen> | the message shows up only on certain bugs but not on others. very weird. |
| 23:04 | <zcorpan> | hsivonen: sure, whenever i find another bug :) |
| 23:04 | <annevk> | Hixie, I was sort of intrigued by your hijacking of 19, should NETWORK_ERR & ABORT_ERR use 20 and 21 instead? |
| 23:04 | <Hixie> | sounds good to me |
| 23:04 | <Hixie> | i just used 19 because it was the first free one after 18 |
| 23:05 | <annevk> | right |
| 23:05 | <Hixie> | and i hijacked 18 cos it was the first free one after 17 |
| 23:05 | <Hixie> | i updated the wiki exception codes page |
| 23:05 | <jacobolus> | annevk: http://www.w3.org/TR/css3-text/#hyphenate has said since the beginning of march: “The WG plans to discuss it at our next face-to-face meeting at the end of March.” |
| 23:05 | <Hixie> | as far as i'm concerned, that's normative http://wiki.whatwg.org/wiki/Exception_Codes :-) |
| 23:05 | <annevk> | I see |
| 23:06 | <jacobolus> | (actually, since march 2007!) |
| 23:06 | <annevk> | jacobolus, heh, hyphenation is difficult and the editors are working on some other stuff too, iirc |
| 23:06 | <Hixie> | the csswg is so dysfunctional |
| 23:06 | <Hixie> | they try to do everything by consensus |
| 23:06 | <Hixie> | and the keep reopening old issues |
| 23:07 | <Hixie> | etc |
| 23:07 | <zcorpan> | why change the NETWORK_ERR and ABORT_ERR codes now? |
| 23:07 | <zcorpan> | (are they implemented?) |
| 23:07 | <Hixie> | if they are we shouldn't change them |
| 23:07 | <Hixie> | it's not a big deal |
| 23:07 | <jacobolus> | if hyphenation worked in browsers (or even just webkit/gecko), the web would be so so so much nicer |
| 23:07 | <jacobolus> | text could actually be justified |
| 23:07 | <jacobolus> | and not look like crap |
| 23:07 | <jacobolus> | s/be/look/ |
| 23:07 | <annevk> | zcorpan, if WebKit ships with them we won't |
| 23:07 | <Hixie> | jacobolus: there are some issues that i would consider to be a much higher priority that are languishing as much :-) |
| 23:08 | <annevk> | zcorpan, if WebKit doesn't it seems better to change them for consistency as 101 and 102 are sort of out of the blue |
| 23:08 | <Hixie> | e.g. getting a decent box model for ui |
| 23:08 | <zcorpan> | annevk: L&S was before, no? |
| 23:08 | <annevk> | yeah, UI, transitions, gradients, etc., menus maybe |
| 23:08 | <annevk> | zcorpan, well, that spec should really be killed |
| 23:08 | <jacobolus> | Hixie: sure. i'm self-interested, in that I want to be able to use Webkit for my own experiments w/ trying to make tools for my personal use for reading long documents on a screen. but still :) |
| 23:08 | zcorpan | doesn't really care |
| 23:09 | <annevk> | zcorpan, nobody really cares, but we should try to minimize confusion :) |
| 23:17 | <annevk> | why won't outerHTML work for XML btw? there's no extra problems there, right? |
| 23:17 | <Hixie> | trying to keep the damage to a minimum |
| 23:17 | <annevk> | i guess you need to do the isHTML check anyway |
| 23:18 | <annevk> | btw, for documentElement.outerHTML, wouldn't documentElement.parentNode.replaceChild(...) do the trick? |
| 23:18 | <Hixie> | the fragment parsing algorithm doesn't handle the root element |
| 23:19 | <Hixie> | you'd have to have a whole other separate codepath for it |
| 23:19 | <zcorpan> | Hixie: what are you going to do with documentFragment? |
| 23:19 | <annevk> | Hixie, ok, fair enough |
| 23:19 | <Hixie> | zcorpan: ? |
| 23:20 | <zcorpan> | Hixie: when the parentNode is a document fragment |
| 23:20 | <Hixie> | oh |
| 23:21 | <Hixie> | uh |
| 23:21 | <Hixie> | i guess we pretend the document element is a <body> |
| 23:22 | <zcorpan> | <zcorpan> opera does the same thing as it does for document |
| 23:22 | <zcorpan> | <zcorpan> webkit throws |
| 23:22 | <zcorpan> | <zcorpan> ie does the same thing as it does for a normal element parent |
| 23:22 | <zcorpan> | * zcorpan likes the ie approach |
| 23:22 | <Hixie> | right |
| 23:23 | <Hixie> | so i guess we pretend the document element is a <body> |
| 23:23 | <Hixie> | er, s/document element/context element/ |
| 23:23 | <Hixie> | matches IE |
| 23:23 | <Hixie> | right? |
| 23:23 | <zcorpan> | yep |
| 23:23 | <Hixie> | eeeeeexcellent |
| 23:23 | <Hixie> | good catch, i totally flaked on that and didn't even think to ask what you mean above |
| 23:24 | <Hixie> | meant |
| 23:28 | Hixie | next looks at insertAdjacentHTML() and shudders |
| 23:30 | <zcorpan> | Hixie: shouldn't execCommand('insertHTML') throw in an xml context? |
| 23:30 | <annevk> | Hixie, how about URL_MISMATCH_ERR becoming 21 and I get 19 and 20 so XHR can define 18-20? |
| 23:30 | <Hixie> | annevk: 18 is going into DOM Core eventually |
| 23:31 | <Hixie> | annevk: i thought webkit implemented 101 102? |
| 23:31 | <annevk> | Hixie, weinig says this is ok for WebKit |
| 23:31 | <weinig> | Hixie: I don't see it as a real compatibility issue |
| 23:31 | <Hixie> | well if you really want to change it then sure *shrug* |
| 23:31 | <annevk> | Hixie, I think NETWORK and ABORT can go into DOM Core as well eventually |
| 23:31 | <Hixie> | feel free to pick whatever numbers you want |
| 23:31 | <Hixie> | update the wiki and i'll adjust as necessary |
| 23:31 | <annevk> | ok |
| 23:31 | <Hixie> | zcorpan: maybe |
| 23:32 | annevk | was just about to offer to edit the wiki |
| 23:32 | <Hixie> | zcorpan: it doesn't use XML, though |
| 23:32 | <weinig> | Hixie: I will change it tonight and see if stuff breaks |
| 23:32 | <weinig> | call it a trial run |
| 23:32 | <Hixie> | k |
| 23:32 | weinig | implemented a dumb version of insertAdjacentHTML recently |
| 23:32 | <zcorpan> | Hixie: otherwise you can inject "malformed" stuff in xml that is so carefully avoided in the rest of the dom |
| 23:33 | <Hixie> | zcorpan: *shrug* |
| 23:33 | <Hixie> | zcorpan: think of it as an editing api |
| 23:33 | <Hixie> | zcorpan: where the user types html without knowing the underlying dom is xml |
| 23:40 | annevk | changed a few normative numbers around :p |
| 23:43 | <weinig> | annevk: one big reason I am not concerned is that I don't think those numbers match IE, do they? |
| 23:43 | <Hixie> | IE doesn't throw DOMException objects |
| 23:44 | <weinig> | Hixie: good point |
| 23:44 | weinig | remembers that discussion fondly now :p |
| 23:44 | <annevk> | you mean the one in Seattle? ... |
| 23:44 | <weinig> | annevk: yes |
| 23:45 | <annevk> | yeah... |
| 23:45 | <weinig> | annevk: is there any reason for non same origin request to not make upload notifications? |
| 23:45 | <Hixie> | hsivonen: "http://www.whatwg.org/specs/web-apps/current-work/source-whatwg":7099.29-7100.28: info warning: Element name "dfnode" cannot be represented as XML 1.0. |
| 23:46 | <Hixie> | hsivonen: seems that error message is bogus (the element name is unknown for sure, but it's fine in xml) |
| 23:47 | <annevk> | weinig, I suppose at that point the preflight check already happened and it's ok |
| 23:48 | <weinig> | annevk: ok, good |
| 23:48 | <annevk> | weinig, even better, now we dropped the PI the same might go for readyState == 2 |
| 23:48 | weinig | nods |
| 23:49 | weinig | eek, I think webkit already does that |
| 23:49 | weinig | did not read close enough |
| 23:50 | <zcorpan> | nn |
| 23:51 | <zcorpan> | Hixie: just one thing: "Note: These are the only ways to make a DOM unserializable." is false given execCommand('insertHTML') :P |
| 23:52 | <annevk> | weinig, given that there's no processing instruction I don't think you're in much trouble |
| 23:52 | <Hixie> | zcorpan: hmm |
| 23:52 | <Hixie> | zcorpan: send mail, something just came up here and i have to run |
| 23:52 | <annevk> | weinig, I'm writing an XMLHttpRequest Level 2 comment e-mail now to remind myself to fix this |
| 23:52 | <Hixie> | zcorpan: but thanks! |
| 23:52 | <Hixie> | a lot :-) |
| 23:53 | <zcorpan> | i'll file a bug |
| 23:53 | <Hixie> | cool |
| 23:53 | Hixie | goes |
| 23:58 | <annevk> | roc, "which probably tells people immediately that I'm an annoying freak, but I think it's good to be open about that." :p |