| 00:00 | <a-ja> | speaking of w3c validator....think its handling of script documentation comments is fubar |
| 00:27 | <Hixie> | well, i've found where my e-mail is going |
| 00:27 | <Hixie> | some sort of spam filter is moving them into a spam folder |
| 00:27 | <Hixie> | but none of my spam filters are configured to move spam there |
| 00:27 | <Hixie> | so... wtf |
| 00:27 | <Philip`> | A spam filter moving spam to a spam folder? That's crazy |
| 00:32 | <Dashiva> | Philip`: The crazy part is that some of the mail makes it through |
| 00:33 | <othermaciej> | yeah I don't care that much about what you call a "resource" - in WebKit internals we don't really draw any distinction between the resource and the resource representation (for example you use a ResourceHandle to talk to a resource, and cache the result as a CachedResource) |
| 00:33 | <TabAtkins> | Yeah, they'll have to fix that. |
| 00:33 | <TabAtkins> | Hixie shouldn't be receiving any email. |
| 00:33 | <othermaciej> | I only got involved because it makes me worry that we'll be buried in word choice bikesheds |
| 00:33 | <TabAtkins> | You worry? |
| 00:33 | <Hixie> | it tells you how much e-mail i receive that about 1500 e-mails got filtered to a spam folder over a period of about a month, 6 weeks, and it took me that long to notice |
| 00:33 | TabAtkins | thought we were already there. |
| 00:34 | <othermaciej> | well, ones that are tied to an ISSUE and therefore require the WG and particularly the chairs to expend some effort to resolve |
| 00:34 | <TabAtkins> | Ah, makes sense. Yes, those are worrying. |
| 00:51 | <TabAtkins> | Hixie, can I talk privately for a sec? |
| 03:02 | <boblet> | I want to display a code sample to copy & paste in something like a form textarea. Given it’s not going to be submitted, is using textarea wrong? |
| 03:08 | <Hixie> | whether it's submitted or not doesn't really matter |
| 03:11 | <othermaciej> | <textarea> will have the helpful effect that Select All will select only its contents |
| 03:11 | <othermaciej> | dunno if other ways of making a scrollable area do that |
| 03:11 | <othermaciej> | (I guess an iframe does but that requires either an external resource or script to fill) |
| 04:23 | <boblet> | Hixie, othermaciej: thanks! |
| 08:12 | <annevk42> | Hixie, you should read my specs more, a bunch of that "other stuff" on the wiki is covered by CSSOM View |
| 08:12 | <Hixie> | cool |
| 08:12 | <Hixie> | i just pasted it in without really looking to be honest |
| 08:12 | <Hixie> | it comes from the XXX markers at the bottom of html5 |
| 08:13 | <Hixie> | most were written in the WF2 days |
| 08:17 | <annevk42> | ah yeah, I remember reading that |
| 08:26 | <Hixie> | http://pimpmyspec.net/ seems to be down |
| 08:26 | <Hixie> | that's gonna put a crimp in my style |
| 08:32 | <annevk42> | whoa, bug count went up to twenty from minus one? |
| 08:32 | <annevk42> | pretty sure I didn't file twenty bugs |
| 08:35 | <Hixie> | -1? |
| 08:36 | <annevk42> | your issue tracker page displayed that |
| 08:36 | <annevk42> | I think because the bug column was empty |
| 08:38 | <Hixie> | oh right, it'll show that when bugzilla is down, which it was |
| 08:42 | <Hixie> | jgraham! |
| 08:42 | <Hixie> | http://pimpmyspec.net/ is down |
| 08:43 | <Hixie> | oh it works now |
| 08:43 | <Hixie> | never mind! |
| 08:53 | <annevk42> | Hixie, where you talk about the default action of the event you should not use "must" at all |
| 08:53 | <annevk42> | Hixie, the algorithm is already required to be followed |
| 08:54 | <hsivonen> | we have quite a few WONTFIX bugs in the W3C bugzilla. those are all potential escalation material :-( |
| 08:54 | <hsivonen> | I hope the chairs won't make the WG debate 96 escalated ISSUEs |
| 08:55 | <annevk42> | hsivonen, looking forward to TPAC already |
| 09:15 | <jgraham> | Hixie: Yeah it seems that the server died or something |
| 09:15 | <Hixie> | annevk2: it is? |
| 09:28 | <annevk2> | Hixie, yeah, from other parts in the spec |
| 09:41 | <zcorpan_> | hmm... window.window=1;alert(window.window) // [object Window] in opera, webkit and firefox |
| 09:41 | <zcorpan_> | window.self=1;alert(window.self) // 1 in webkit and firefox, [object Window] in opera |
| 09:41 | <zcorpan_> | html5 just says readonly attribute |
| 09:42 | <zcorpan_> | so i guess according to html5, both should throw on setting? |
| 09:42 | <zcorpan_> | what happens in ie? |
| 09:44 | <zcorpan_> | and what should happen in a WorkerGlobalScope when setting self.self ? |
| 09:46 | Philip` | thought readonly was just ignored on setting |
| 09:46 | <Philip`> | (rather than throwing) |
| 09:46 | <zcorpan_> | "It is language binding specific whether assignment is simply disallowed by the language, ignored or an exception is thrown." |
| 09:47 | <Philip`> | Where is the language binding specified? |
| 09:49 | <zcorpan_> | "If the attribute is declared readonly, the property has attributes { DontDelete, ReadOnly }. Otherwise, the property has attributes { DontDelete }." |
| 09:50 | <zcorpan_> | ReadOnly in javascript means nothing happens on setting, right? |
| 09:50 | <zcorpan_> | i guess firefox and webkit have [Replacable] for window.self |
| 09:51 | Philip` | tries to look at ES5, but can't work out what it's saying |
| 09:52 | <Philip`> | *That* is a spec designed solely for implementors, not users |
| 09:52 | <zcorpan_> | "when the ReadOnly attribute for a property is set to true, any attempt by executed |
| 09:52 | <zcorpan_> | ECMAScript code tochange the value of thepropertyhas noeffect." says ecma262 |
| 09:52 | <zcorpan_> | though that might be non-normative... |
| 09:53 | <Philip`> | ES5 says "any attempt by executed ECMAScript code to change the value of the property fails", which is less helpful because I don't know what "fails" means |
| 09:53 | <Philip`> | (Also it probably varies in strict mode, whatever that is) |
| 09:54 | <hsivonen> | what's the purpose of the ES5 "strict mode"? |
| 09:54 | <Philip`> | Helping programmers write code with fewer bugs? |
| 09:54 | <Philip`> | like a lint tool but at runtime |
| 09:54 | <hsivonen> | is the non-strict mode of ES5 compatible with existing ES3 scripts? |
| 09:54 | <zcorpan_> | "The property is a read-only property. Attempts by ECMAScript code to write to |
| 09:54 | <zcorpan_> | the property will be ignored." ... "An i mpl ement at i on of ECMAScr i pt must behave as i f i t pr oduced and |
| 09:54 | <zcorpan_> | oper at ed upon i nt er nal pr oper t i es i n t he manner descr i bed her e." ...is probably the normative part |
| 09:55 | <jgraham> | In ES5 [[Put]] can either throw or not throw |
| 09:55 | <Philip`> | Yay PDF copy-and-paste |
| 09:55 | <zcorpan_> | Philip`: yeah |
| 09:55 | <jgraham> | IIRC it is called with throw=false for the normal case of property setting |
| 09:55 | <hsivonen> | Philip`: If I have an ES3 browser and I say "use strict" and write non-strict code, will my code run on the ES3 browser but fail in an ES5 browser? |
| 09:55 | <hsivonen> | or will the ES5 browser just warn to console or something? |
| 09:56 | <zcorpan_> | hsivonen: it will fail |
| 09:56 | <zcorpan_> | hsivonen: some things throw in strict mode |
| 09:56 | <zcorpan_> | hsivonen: like arguments.caller |
| 09:56 | <hsivonen> | zcorpan_: but won't fail in ES3? |
| 09:56 | <zcorpan_> | right |
| 09:57 | Philip` | uses JSOPTION_STRICT when embedding SpiderMonkey in non-web things, because it helps detect some bugs |
| 09:57 | <hsivonen> | seems like a disaster if people start using "use strict" as a shiny thing before upgrading to an ES5 browser |
| 09:57 | <Philip`> | (but I think that option only results in warnings, never changes to behaviour) |
| 09:57 | <zcorpan_> | hsivonen: indeed |
| 09:57 | <jgraham> | It seems to very with strict mode |
| 09:58 | <jgraham> | See the PutValue algorithm for the normative description |
| 09:58 | <zcorpan_> | hsivonen: although apparently "use strict" allows for some optimizations |
| 09:58 | <hsivonen> | zcorpan_: ok. that makes more sense |
| 09:59 | <jgraham> | hsivonen: ES5 is not absolutely backwards compatible with ES3 |
| 09:59 | <jgraham> | even in non-strict mode |
| 09:59 | <hsivonen> | jgraham: are browsers expected to ship only an ES5 engine or to sniff between ES3 and ES5 behavior somehow? |
| 10:00 | <Hixie> | zcorpan_: if i need to add [Replaceable] to 'self' let me know |
| 10:00 | <jgraham> | Ship only an ES5 engine |
| 10:00 | <hsivonen> | jgraham: I see |
| 10:00 | <jgraham> | (the differences are things like regexp literals evaluating to different objects rather than always the same object) |
| 10:01 | <jgraham> | (which would can observe through the lastIndex property) |
| 10:01 | <zcorpan_> | http://www.google.com/codesearch?q=self+lang%3Ajs&hl=en&btnG=Search+Code |
| 10:01 | <zcorpan_> | hmm, first hit has var self = arguments.callee :( |
| 10:01 | <jgraham> | s/would/you/ |
| 10:03 | <zcorpan_> | hmm but that works in opera |
| 10:03 | <jgraham> | AFAICT the main motivators for strict mode are 1) a desire to make more substantial language cleanups then you can make without breaking backward compatibility b) a desire to have a language better suited to Caja and other such security subsets 3) Maybe some potential for optimization |
| 10:03 | <jgraham> | s/b/2 |
| 10:04 | <Philip`> | Does ES5 not have any of the new keywords that JS1.n (n >= 6) added? |
| 10:04 | <jgraham> | iirc it hasn't added any new keywords yet |
| 10:04 | <jgraham> | They keep talking about it for ES6 though |
| 10:05 | <zcorpan_> | what's the plan to not break compat in ES6? |
| 10:06 | <jgraham> | Well theoretically most of the new keywords are future reserved words. But I don't know if they are in the intersection of the future reserved words that are actually reserved by all browsers |
| 10:06 | <jgraham> | Or even in the union |
| 10:06 | <zcorpan_> | yield isn't a reserved word in browsers, right? |
| 10:07 | <Philip`> | 'let' and 'yield' aren't listed as future reserved words in ES5 |
| 10:07 | <Philip`> | (at least the draft I'm looking at) |
| 10:07 | <Philip`> | but it says "The identifiers ‗let‘, and ‗yield‘ may be used in a future version of this standard. |
| 10:07 | <Philip`> | " |
| 10:07 | <jgraham> | Oh. Well that shows how useful that list was |
| 10:08 | <hsivonen> | Philip`: are those used in anything but Firefox chrome? |
| 10:08 | <jgraham> | (I only remember enough about the list of future reserved words to remember that it is basically fiction) |
| 10:09 | <jgraham> | Oh, let and yield are future reserved words in strict mode only |
| 10:09 | <jgraham> | ]and they seem to have cut down the list in non-strict mode since I last checked |
| 10:09 | <Philip`> | hsivonen: http://canvex.lazyilluminati.com/3dmodel/example0.html uses 'yield' |
| 10:10 | <Philip`> | Doubt it's used much in real content, though |
| 10:10 | <hsivonen> | Philip`: are they supported by engines other than SpiderMonkey? |
| 10:11 | <Philip`> | hsivonen: No, as far as I'm aware |
| 10:14 | <zcorpan_> | are TC39 aware of the Web EcmaScript wiki page? |
| 10:14 | <Philip`> | http://www.kpmg.de/Themen/7771.htm is the only page I see with version=1.[6789] in <script type> |
| 10:14 | <jgraham> | zcorpan_: Not as far as I'm aware |
| 10:15 | <zcorpan_> | maybe someone should bring their attention to it, or at least to the issues it tries to cover |
| 10:17 | jgraham | cannot imagine anything good soming of it |
| 10:17 | <jgraham> | *coming |
| 10:17 | <hsivonen> | Philip`: and it seems to use it only for the purpose of recording the version recognized by the browser for tracking purposes |
| 10:18 | <zcorpan_> | Philip`: that page just checks that version is not 1.0 or 1.1 |
| 10:21 | <Philip`> | Indeed |
| 10:24 | <annevk2> | Hixie, I also filed another bug about fallback and online whitelist namespace being matched by something more complicated than just a prefix match |
| 10:24 | <hsivonen> | are ES5 browsers supposed to modify behavior on version=1.1 or version=1.2, etc.? |
| 10:25 | <hsivonen> | what about e4x=1? |
| 10:25 | <zcorpan_> | hsivonen: i would guess that opera will modify behavior of version=1.1 and not support e4x |
| 10:27 | <hsivonen> | zcorpan_: I see. Does ES5 specify what the delta is when version=1.1 is specified? |
| 10:28 | <zcorpan_> | don't know |
| 10:28 | <zcorpan_> | but i doubt it :) |
| 10:28 | <zcorpan_> | it probably says what the delta is compared to ecma262 |
| 10:28 | <jgraham> | ES5 says nothing about any of this |
| 10:29 | <hsivonen> | jgraham: is there a spec that specifies this part of the interoperable platform? |
| 10:30 | <jgraham> | hsivonen: Not unless HTML5 says something about it |
| 10:31 | <hsivonen> | hmmkay |
| 10:31 | <jgraham> | In general ES5 is silent about how scripts are loaded |
| 10:31 | <hsivonen> | is version=1.1 truly version 1.1 or the latest plus a specific delta? |
| 10:31 | <Philip`> | I didn't think any current browsers (except Firefox for versions >= 1.6) changed behaviour on version at all |
| 10:32 | <zcorpan_> | in opera version=1.1 is the same as version=1.5, i think |
| 10:33 | <Philip`> | See https://bugzilla.mozilla.org/attachment.cgi?id=167555 and/or http://philip.html5.org/demos/js/jstype.html |
| 10:33 | <Philip`> | Oh, don't bother with the latter one, it doesn't show old version numbers |
| 10:35 | Philip` | tweaks |
| 10:35 | <Philip`> | Now it does |
| 10:49 | <Hixie> | annevk2: what does http://www.w3.org/Bugs/Public/show_bug.cgi?id=7760 mean? |
| 10:51 | <annevk2> | Hixie, the summary table suggests error is the last event you get while after that noupdate is dispatched |
| 10:51 | <annevk2> | though I later realized that maybe noupdate was not dispatched to the same place as where error was dispatched, but did not check |
| 10:52 | <annevk2> | I also think it would be helpful to have separate event summaries for different scenarios |
| 10:52 | <zcorpan_> | which is right? regexp matches string, or string matches regexp? |
| 10:52 | <Hixie> | right, it either sends noupdate or error |
| 10:53 | <Hixie> | trying to describe summaries for different scenarios is reallllly hard |
| 10:53 | <Hixie> | because there's a bazillion different scenarios |
| 10:53 | <Philip`> | zcorpan_: Depends what your favourite programming language does |
| 10:53 | <Hixie> | i had that problem with drag and drop too |
| 10:53 | <annevk2> | the multi-window scenario is most interesting I suppose; where the other window is doing the update but you're receiving events too, but less |
| 10:54 | <Philip`> | $str =~ /regexp/ vs regexp.match(str) etc |
| 10:54 | <Philip`> | zcorpan_: (I don't think it matters much in practice since it's unambiguous either way) |
| 10:55 | <annevk2> | because if you split those it is more clear why noupdate can come directly after checking without downloading |
| 10:55 | <Hixie> | annevk2: sounds like something lachy should document in his doc |
| 10:56 | <annevk2> | devil's advocate, but shouldn't the authoring details be just as clear as the impl details? |
| 10:56 | <zcorpan_> | Philip`: yeah. i guess different people have different mental models |
| 10:58 | <annevk2> | in general it seems that event details are not very author friendly |
| 10:59 | <annevk2> | e.g. dispatching of online and offline events is impl-protected |
| 10:59 | <zcorpan_> | i think all elements and objects should have an author box saying which events are fired and when |
| 11:00 | <zcorpan_> | video has it |
| 11:00 | <zcorpan_> | but other things don't |
| 11:01 | <Hixie> | annevk2: every time i've written detailed information for the author side that is just informative, i've regretted it. So I'd really rather not include any tutorial-like stuff in the spec until at least mid-CR. |
| 11:02 | <annevk2> | if you want to do it eventually then it works for me |
| 11:02 | <annevk2> | i agree that such authoring fluff is annoying; I haven't done it for XHR either |
| 11:03 | <annevk2> | though I have to say that appcache is a bitch to review |
| 11:04 | <annevk2> | Hixie, is it too late to change names of appcache events? |
| 11:06 | <Hixie> | why would we change the names? |
| 11:07 | <Hixie> | for http://www.w3.org/Bugs/Public/show_bug.cgi?id=7769 btw do you mean align on deletion, or on making the content attribute text be updated when the IDL attribute is updated? |
| 11:07 | <annevk2> | make the content attribute updated; because otherwise deletion wouldn't work |
| 11:08 | <Hixie> | what should it be set to if the event handler is set to a java method? |
| 11:08 | <annevk2> | so with cached it's clear it ends a phase; but obsolete doesn't match that; it would have to be obsoleted, notupdated, etc. |
| 11:09 | <annevk2> | Hixie, I'm fine with restricting this to ECMAScript |
| 11:09 | <annevk2> | in fact, I'd be fine with not exposing on* in any other scripting language |
| 11:09 | <Hixie> | annevk2: file a bug, i guess. i dunno who has implemented the events. |
| 11:10 | <Hixie> | annevk2: so what should onfoo="" be set to if onfoo is set to function () { } ? |
| 11:10 | <annevk2> | every function has a toString() no? |
| 11:11 | jgraham | notes that function.toString() can be overridden |
| 11:12 | <Hixie> | annevk2: so what if toString() throws an exception? |
| 11:12 | <zcorpan_> | annevk2: lunch |
| 11:13 | <annevk2> | Hixie, not sure :/ |
| 11:13 | <annevk2> | Hixie, if you want me to figure it out more I can have a look |
| 11:13 | <Hixie> | http://software.hixie.ch/utilities/js/live-dom-viewer/saved/249 suggests your description of firefox is wrong |
| 11:14 | <annevk2> | zcorpan_, you got it |
| 11:14 | <annevk2> | Hixie, yeah, I couldn't reproduce what hallvord told me either; I'll look into it some more and update the bug report |
| 11:14 | <annevk2> | back in a bit |
| 11:27 | <othermaciej> | does anyone here know PHP? |
| 11:27 | <othermaciej> | (and would forgive me for asking a dumb question?) |
| 11:27 | hsivonen | used to know PHP but has avoided it for the last 5 or so years |
| 11:28 | <othermaciej> | how do you get the HTTP request body? |
| 11:28 | <othermaciej> | the script I am trying to modify gets the request method thus: $_SERVER['REQUEST_METHOD'] |
| 11:29 | <hsivonen> | I don't know/recall. I always used facilities that didn't require me to access the request body directly and parsed it for me. |
| 11:31 | <Hixie> | the request body in CGI scripts is text on stdin |
| 11:31 | <Hixie> | dunno how that helps in php |
| 11:32 | <othermaciej> | I could get it if I could read stdin I guess |
| 11:33 | <Philip`> | I don't think PHP typically runs as CGI |
| 11:34 | <Philip`> | since it uses mod_php instead |
| 11:34 | <jgraham> | $request_body = @file_get_contents('php://input'); according to google |
| 11:35 | <jgraham> | or maybe http_get_request_body() if you have the right extension module installed |
| 11:35 | <Philip`> | http://www.daniweb.com/code/snippet216846.html does that |
| 11:36 | hsivonen | is amused by the IANA-incompliant URI used to address an HTTP *request* body |
| 11:36 | <Philip`> | I guess it's conceptually equivalent to reading STDIN |
| 11:36 | <Philip`> | rather than being intended for HTTP request bodies in particular |
| 11:37 | <hsivonen> | Philip`: sure but still |
| 11:37 | hsivonen | is easily amused |
| 11:37 | <Philip`> | Oh, I guess wrong |
| 11:37 | <Philip`> | php://stdin is for stdin |
| 11:38 | <Philip`> | 'php://input allows you to read raw POST data. It is a less memory intensive alternative to $HTTP_RAW_POST_DATA and does not need any special php.ini directives. php://input is not available with enctype="multipart/form-data".' |
| 11:38 | <Philip`> | I guess you could use $HTTP_RAW_POST_DATA, then |
| 11:38 | <Philip`> | Oh, except for the php.ini stuff |
| 11:38 | <othermaciej> | I want the raw post data and I don't care about memory, this is for a regression test |
| 11:39 | <Philip`> | readfile("php://filter/read=string.toupper|string.rot13/resource=http://www.example.com";); |
| 11:39 | <Philip`> | That looks fun |
| 11:41 | <Dashiva> | I never understood why PHP was so restrictive about raw post data on regular form submits |
| 11:41 | <Philip`> | Maybe because PHP itself automatically reads all the data and outputs it to files, and then the raw post data is no longer available because it's all been read |
| 11:42 | <Philip`> | (and they don't want to buffer all the megabytes of file uploads just in case somebody might want to access the post data again later) |
| 11:42 | <othermaciej> | none one php://input, php:/stdin or $HTTP_RAW_POST_DATA seems to work |
| 11:43 | <othermaciej> | (reading stdin hangs, other two seem to get nothing) |
| 11:43 | <othermaciej> | tcpdump confirms I an indeed sending an http body |
| 11:43 | <Philip`> | What content-type are you sending? |
| 11:44 | <othermaciej> | ah, I seem to be dropping the request Content-Type |
| 11:45 | <othermaciej> | that's probably my bug |
| 12:02 | <othermaciej> | OK, I'm now sending Content-Type: application/x-www-form-urlencoded |
| 12:02 | <othermaciej> | a non-PHP test case works but this PHP one still does not |
| 12:02 | <othermaciej> | I am doing <?php @file_get_contents('php://input') ?> |
| 12:03 | <Hixie> | well i was hoping to do microdata tonight |
| 12:03 | <hsivonen> | does PHP always intercept application/x-www-form-urlencode ? |
| 12:03 | <Hixie> | but i'm getting too tired to work on it |
| 12:03 | <Dashiva> | othermaciej: Isn't that supposed to work only when you _don't_ use x-www-form-urlencoded? |
| 12:03 | <Hixie> | so i'll go to bed instead |
| 12:03 | <Hixie> | nn |
| 12:04 | <hsivonen> | nn |
| 12:04 | <othermaciej> | Dashiva: I think what it does not like is "multipart/form-data" |
| 12:05 | <Dashiva> | Hum |
| 12:06 | <Philip`> | othermaciej: Don't you need to use 'print' or 'echo' or something? |
| 12:07 | <othermaciej> | Philip`: ah |
| 12:07 | <Philip`> | Also, what does the '@' do? |
| 12:08 | <Dashiva> | No error reporting |
| 12:08 | <Philip`> | Oh |
| 12:08 | <Philip`> | I guess you don't want that |
| 12:09 | <zcorpan_> | doesn't php require a semicolon? |
| 12:10 | <othermaciej> | Philip`: thanks for all your help |
| 12:11 | <Philip`> | Of course what you should do is replace all the PHP with Perl |
| 12:11 | <Philip`> | print "Content-type: text/plain\n\n", <>; or whatever |
| 12:12 | <othermaciej> | I should, but that's beyond the scope of this bug fix |
| 12:12 | <othermaciej> | now that I did this I can get back to implementing HTML5 elements (tomorrow) |
| 12:25 | <Lachy> | Wondering if I should just give in to the requests to change :reference back to :scope as many people are requesting and live with the less than ideal name? |
| 12:26 | <zcorpan_> | :reference is a bit long |
| 12:26 | <Dashiva> | Is there a third name that would serve as a compromise that annoys everyone equally? |
| 12:27 | <zcorpan_> | :ref is shorter |
| 12:27 | <zcorpan_> | but i'd be happy with :scope too |
| 12:27 | <Lachy> | I once called it :context, but people didn't like htat |
| 12:27 | <Lachy> | *that |
| 12:28 | <jgraham> | Dashiva: :lachyForKing would annoy almost everyone equally |
| 12:28 | <zcorpan_> | what's wrong with :scope? |
| 12:28 | <Lachy> | I considered :ref, but all other pseudo-classes in use full, non-abbreviated words, so I decided against it for consistency |
| 12:29 | <Lachy> | zcorpan_, it's only a semi-accurate name for a limited set of the use cases. |
| 12:30 | <Lachy> | since the elements it matches don't necessarily define any sort of scope. |
| 12:30 | <Lachy> | although it does in scoped stylesheets and queryScopedSelector, it doesn't with matchesSelector or with querySelectorAll("...", refNodes); |
| 12:31 | <zcorpan_> | there's a queryScopedSelector? |
| 12:31 | <Lachy> | there is now |
| 12:31 | <Lachy> | http://dev.w3.org/2006/webapi/selectors-api2/ |
| 12:35 | <zcorpan_> | Lachy: doesn't web idl use in optional foo bar instead of [, foo bar] ? |
| 12:35 | <zcorpan_> | Lachy: also, shouldn't "Any" be lowercase? |
| 12:37 | <zcorpan_> | are :this or :self bad names? |
| 12:37 | <Dashiva> | I'd say so |
| 12:38 | <zcorpan_> | why |
| 12:38 | <Lachy> | because there's more than one and it doesn't necessarily match the context node |
| 12:38 | <Dashiva> | Scoped stylesheets |
| 12:39 | <Philip`> | Have two different names which have exactly the same effect, then people can choose whichever one fits best |
| 12:39 | <Lachy> | all the names previously considered, that I can remember are: :scope, :this, :self, :context, :reference, :context-node |
| 12:40 | <Dashiva> | :center-of-attention |
| 12:41 | <Lachy> | Dashiva, it's a little short, don't you think? |
| 12:41 | <Dashiva> | Wait, I got it |
| 12:41 | <Dashiva> | :important |
| 12:44 | <Philip`> | :! |
| 12:44 | <zcorpan_> | :: |
| 12:45 | <Philip`> | :D |
| 12:45 | <zcorpan_> | :lol: |
| 12:45 | <Lachy> | hah |
| 12:45 | <Lachy> | that would be cool having an emoticon as a pseudo-class. |
| 12:45 | <Dashiva> | :S for scope |
| 12:46 | <Philip`> | Selectors are too dull and unemotional |
| 12:46 | <Philip`> | If all the symbols were replaced with emoticons, you'd have an exciting rollercoaster ride of feeling as you read the selector |
| 12:46 | <Lachy> | :p would be best, if we could think of something for it to stand for |
| 12:47 | <Philip`> | It should match paragraphs |
| 12:47 | <Philip`> | (in the HTML5-defined sense, not just <p>s) |
| 12:49 | <Lachy> | :-^)> |
| 12:50 | <Lachy> | :-^)>-< |
| 13:00 | <Lachy> | http://dev.w3.org/2006/webapi/WebIDL/#idl-any |
| 13:00 | <Lachy> | It says 'Its type name is “Any”. ' But then in the grammar http://dev.w3.org/2006/webapi/WebIDL/#dfn-type-name it uses "any" |
| 13:00 | <Lachy> | I'm confused. |
| 13:01 | <Lachy> | heycam, ^ |
| 13:16 | <hsivonen> | http://blog.piraattipuolue.fi/2009/09/sananvapaustuomio-tekijanoikeuden-varjolla/ |
| 13:16 | <hsivonen> | oops. wrong window |
| 13:17 | <hsivonen> | (summary: parody that made a politically relevant point got a sentence of copyright infringement to its creator) |
| 13:30 | <remysharp> | I assume someone knows that the multipage spec url doesn't work atm: http://www.whatwg.org/specs/web-apps/current-work/multipage/ |
| 13:30 | <hsivonen> | maybe related to the PMS problem earlier today? |
| 13:32 | <remysharp> | Erm, not sure if that's a joke or serious, but the multi page was a useful resource :) |
| 13:32 | remysharp | hopes it comes back to life |
| 13:33 | <hsivonen> | remysharp: PMS is pimpmyspec.net that Hixie uses to generate the spec from the source file |
| 13:33 | <remysharp> | nice! |
| 13:34 | <remysharp> | my old company, we managed to produce a system called: Multi Infrastructure List Framework |
| 13:35 | <remysharp> | it took 6 months before the CTO flagged up the documentation - by which point it was way too late to change all the references to MILF |
| 13:56 | <zcorpan_> | opera has a framework called WTF |
| 13:56 | <zcorpan_> | (widget testing framework) |
| 13:57 | <remysharp> | there needs to be a wiki for those sneaky things that get in. |
| 13:57 | <remysharp> | WTF is superb. |
| 13:57 | <Rik|work> | zcorpan_: webkit has WTF too |
| 13:57 | <Rik|work> | Web Template Framework IIRC |
| 14:11 | <Philip`> | http://www.whatwg.org/specs/web-apps/current-work/multipage/stderr.txt |
| 14:11 | <Philip`> | I hope that's not my bug |
| 14:12 | <Philip`> | but my machine that runs the splitter seems to be currently offline so I can't do much about it now :-( |
| 14:12 | <Philip`> | (I don't see how it'd get that error unless some id was in the TOC but not in the spec, which should be impossible unless the spec was truncated) |
| 14:22 | <annevk2> | hmm, the bug list is back to zeor |
| 14:22 | <annevk2> | time to file new bugs |
| 14:24 | hsivonen | wishes the www.cnn.com came with minimized test cases that hit all the same bugs only once |
| 14:24 | <hsivonen> | s/wishes the/wishes/ |
| 14:24 | <jgraham> | hsivonen: Debugging real world websites is fun, no? |
| 14:25 | <hsivonen> | "fun" |
| 14:25 | <jgraham> | Yeah, scare quotes implied |
| 14:26 | jgraham | wonders if anyone has written a tool for nicely formatting javascript |
| 14:27 | <jgraham> | Particularly to use on stuff that has been minimized |
| 14:27 | <Philip`> | Firefox has |
| 14:27 | <Philip`> | uneval() |
| 14:27 | <Philip`> | (Wrap the code in a function and then uneval the function, in particular) |
| 14:27 | jgraham | also wonders how much js-minimization actually helps given gzip and caching |
| 14:28 | <jgraham> | Philip`: Ah, should have thought of that |
| 14:28 | <hallvors> | jgraham: there are a number of tools for formatting JS |
| 14:29 | <hallvors> | http://hallvord.com/opera/scriptformatter.php |
| 14:29 | <hallvors> | http://jsbeautifier.org/ |
| 14:29 | <Philip`> | jgraham: Seems like a very easy thing to test |
| 14:29 | <hallvors> | http://elfz.laacz.lv/beautify/ |
| 14:29 | <hallvors> | I mostly use the first one :-p |
| 14:29 | <jgraham> | Philip`: The caching part is harder to test |
| 14:30 | <hsivonen> | so... I'm getting a document.write() that implies document.open() and it's not even a defer script |
| 14:30 | <hsivonen> | hmm. |
| 14:30 | <gsnedders> | The latter two both run in JS, and are rather slow on, e.g., Gmail :P |
| 14:30 | <Philip`> | jgraham: On an ancient version of jQuery, minimisation saves ~50% on the gzipped size |
| 14:31 | <jgraham> | hallvors: Nice. I will remember those when the unenviable day comes that I have to grapple with a large minimized script |
| 14:31 | <hallvors> | (yeah, given all the string manipulation you need to do I'm not surprised that all of them probably are slow) |
| 14:31 | <gsnedders> | hallvors: It'd be nice to use a C++ tokenizer and work from that… but for me to write that I'd have to learn C++ :P |
| 14:31 | <jgraham> | Philip`: What about if you strip comments only from the unminimized version |
| 14:31 | <hallvors> | in Opera, viewing source of the script, taking it through the formatter, and re-loading from cache gives you an entirely new debugging experience |
| 14:31 | Philip` | hopes the "string manipulation" consists of something like a parser, not just a load of regexps |
| 14:32 | jgraham | is always somewhat nbervous of that |
| 14:32 | <hallvors> | Philip`: mine does. Can't speak for the others, though I think the elfz.laacz.lv one is quite similar to mine |
| 14:32 | <hallvors> | jsbeautifier.org has bugs.. It tends to break complex scripts. |
| 14:33 | <gsnedders> | hallvors: Until you hit an inline script in a page that cannot be cached, which means you can't edit it and use Dragonfly. |
| 14:33 | <hsivonen> | hallvors: thanks. Those could have been useful when figuring out the MySpace XMLSerializer silliness |
| 14:33 | <hallvors> | gsnedders: in that case, use the user JS equivalent :) |
| 14:33 | <Philip`> | jgraham: Original .js is about 26KB (gzipped), packed is 14KB, non-packed but removing comments (in particular just running it through 'cpp') is 17KB |
| 14:33 | <hallvors> | http://hallvord.com/opera/scriptformatter.js |
| 14:33 | <gsnedders> | hallvors: That's a pain :) |
| 14:34 | <hallvors> | why? :) |
| 14:34 | <hsivonen> | http://jsbeautifier.org/ doesn't work with an HTML5 parser, so that one is problematic for me |
| 14:34 | <gsnedders> | hallvors: Means editing another file :) |
| 14:34 | <hallvors> | put scriptformatter.js in your user js folder and Dragonfly will see all source code nicely formatted |
| 14:34 | <gsnedders> | ooo… |
| 14:34 | <jgraham> | Philip`: I would have thought that 3Kb would be pretty unnoticable on most sites |
| 14:35 | gsnedders | hopes he remembers that until he gets back to work on Monday |
| 14:35 | <hallvors> | ..or use Fiddler's autoresponse feature on the formatted source (which is what you have to do to debug formatted code in other browsers anyway) |
| 14:35 | <Philip`> | jgraham: Probably true, but if you're already going through all the effort of stripping comments and having separate development vs deployment versions of your scripts, why not save a few extra KB at the same time? |
| 14:36 | <Philip`> | Also it makes it harder for people to steal your code |
| 14:36 | <gsnedders> | hallvors: Who says I use Windows? :) |
| 14:36 | <jgraham> | Philip`: Because browser makes will hate you? |
| 14:36 | <jgraham> | *makers |
| 14:36 | <hallvors> | hsivonen: you're welcome. Hope you find it useful and do complain about any bugs that break scripts after formatting |
| 14:36 | <Philip`> | jgraham: That's just payback |
| 14:36 | <gsnedders> | jgraham: Browser vendors will hate them for having horrible JS to start with anyway. |
| 14:36 | <Philip`> | They'll be so fed up with stupid browser bugs that they'll be happy to make your lives as miserable as possible |
| 14:36 | <jgraham> | gsnedders: Not all sites have horrible js |
| 14:37 | <gsnedders> | jgraham: Most do. |
| 14:37 | <hallvors> | gsnedders: well, find a windows box to use Fiddler on then, and proxy your traffic through it :-p |
| 14:37 | gsnedders | hopes jgraham doesn't find out what book he's reading at the moment, as then he'll really hate him |
| 14:37 | <jgraham> | What, Twilight? |
| 14:38 | gsnedders | nods |
| 14:38 | <jgraham> | You suck |
| 14:38 | <gsnedders> | Actually, the second one. |
| 14:38 | <Philip`> | You're reading Noddy? |
| 14:38 | <gsnedders> | It's so badly written, but I've fallen in love wit it nevertheless. |
| 14:39 | <gsnedders> | Philip`: No, "New Moon" |
| 14:39 | <Philip`> | Oh, right, you were agreeing with jgraham, not giving a hint |
| 14:40 | gsnedders | finds a photo so over-exposed the face in it is white even turning exposure all the way down on the RAW photo… |
| 14:41 | <jgraham> | gsnedders: YOu only have 12bits of luminance information. If it clips it's gone |
| 14:41 | <gsnedders> | Right, I know. |
| 14:42 | <gsnedders> | I don't think I've ever had it clip on the subject of the photo before, though I have on the sky. |
| 14:42 | <jgraham> | You don't own a flash? |
| 14:43 | <gsnedders> | No |
| 14:43 | <gsnedders> | (The background light for the photo where the face is over-exposed was done in perfectly fine lighting outdoors, a flash wouldn't have helped) |
| 14:43 | <gsnedders> | (I guess the sun came out from behind the clouds at the instant I took it) |
| 14:44 | <gsnedders> | (Everything else I have from those few minutes is fine) |
| 14:46 | <gsnedders> | (Well, apart from some chromatic aberrations in a couple) |
| 14:47 | <jgraham> | No I meant if you had a flash you would be used to overexposed foregrounds |
| 14:47 | <gsnedders> | Ah. |
| 14:47 | <jgraham> | And my point earlier was more that it is usually obvious when things have clipped without adjusting anything in the RAW (indeed the software can tell you) |
| 14:48 | gsnedders | doesn't mention anything else that has happened this month that jgraham will disagree with |
| 14:48 | <hsivonen> | nice. it seems the problem I had was related to running a fragment parser in the context of a document that has an active network stream parse going on |
| 14:49 | <hsivonen> | and the document confused innerHTML with document.write code and ran some document.write finalization code |
| 14:49 | <jgraham> | gsnedders: Now I want to know |
| 14:50 | <gsnedders> | jgraham: You don't think I was teasing you or anything? |
| 14:51 | <jgraham> | No. You would never do that |
| 14:52 | <gsnedders> | jgraham: Because someone downstairs (there) said I was, "a very nice, a bit humble and easy going person"? |
| 14:54 | <jgraham> | I thought that was just to get you an apartment contract :p |
| 14:54 | <gsnedders> | Well, yeah :P |
| 14:58 | <TabAtkins> | Philip`, you break XML-for-real blogs for fun, right? I've got a friend who wants some examples, as he's interested in the theory. |
| 14:58 | <krijn> | Why is http://krijnhoetmer.nl/zooi/html/document-write-js.html not allowed by HTML5? |
| 14:59 | <krijn> | While http://krijnhoetmer.nl/zooi/html/test2.html is |
| 15:09 | <Philip`> | TabAtkins: Yes |
| 15:09 | <Philip`> | "fun" |
| 15:09 | <TabAtkins> | Hehe. |
| 15:11 | <Philip`> | TabAtkins: http://diveintomark.org/archives/2008/03/09/no-fury-like-dracon-scorned#comment-11442 had some links |
| 15:11 | <Philip`> | to examples |
| 15:11 | <Philip`> | Otherwise, read the IRC log to find more examples :-) |
| 15:12 | <TabAtkins> | Philip`, hah. |
| 15:13 | krijn | should scrape the logs for links |
| 15:15 | <Philip`> | TabAtkins: I'm not sure what "the theory" is |
| 15:16 | <TabAtkins> | Dunno, he was a bit vague. |
| 15:16 | <jgraham> | I guess "the theory" is the bits of XML 1.0 + XML NS 1.0 that are most commonly incorrect |
| 15:17 | <jgraham> | (ly implemented) |
| 15:17 | <Philip`> | It basically just involves putting U+0000 and U+FFFF into query strings until something breaks |
| 15:17 | <TabAtkins> | Makes sense. |
| 15:17 | <TabAtkins> | His exact phrasing was "I'd love some examples/references. Breaking websites is fun, and useful now that I'm writing some of my own." |
| 15:18 | <Philip`> | or into comments, if you want to break the page for everybody who looks at it |
| 15:19 | <zcorpan_> | or in trackbacks |
| 15:20 | <Philip`> | Bonus points if you prevent people using the admin interface to clean up your mess |
| 15:20 | <zcorpan_> | /404%00 |
| 15:21 | <TabAtkins> | Yeah, I regaled him with tales (I think I cribbed it from diveintomark) of the draconian-error advocate who got a bad trackback and was unable to fix it because it killed his backend. |
| 15:23 | Philip` | wonders how vulnerable Google Wave will be to XML well-formedness issues |
| 15:23 | <Philip`> | (Someone should give me an invite :-p ) |
| 15:31 | <heycam> | Lachy, use "any" in the idl |
| 15:31 | <Lachy> | heycam, why is the spec confusing? |
| 15:31 | <heycam> | the "type name" is a string used for other purposes |
| 15:31 | <heycam> | maybe i should rename that definition |
| 15:32 | <Lachy> | also, do I need to make the module dom explicit? Can I leave that out of the IDL? |
| 15:33 | <heycam> | no as currently written i believe you can leave out the module dom |
| 15:34 | <Lachy> | ok. I will remove it later |
| 15:39 | <zcorpan_> | why is ErrorEvent's .filename not called .source for consistency with <body onerror>? |
| 15:47 | <aroben> | http://www.whatwg.org/specs/web-apps/current-work/multipage/ seems broken |
| 15:48 | <Philip`> | Known issue |
| 15:48 | <Philip`> | I blame Hixie and/or jgraham |
| 15:48 | <Philip`> | and/or my ISP |
| 15:48 | <Philip`> | so it can't be fixed yet |
| 15:49 | <aroben> | ok, thanks |
| 16:19 | <miketaylr> | just noticed that opera defines a validity object for input elements |
| 16:19 | <miketaylr> | is this documented anywhere, does anyone know? |
| 16:20 | <annevk2> | yes, HTML5 |
| 16:20 | <miketaylr> | :S |
| 16:20 | miketaylr | fails |
| 16:20 | <miketaylr> | thx annevk2 |
| 16:20 | <Dashiva> | I'd link it, but Philip` broke the spec |
| 16:23 | <Philip`> | :'-( |
| 16:24 | <Philip`> | You could link to the single-page version |
| 16:24 | <Philip`> | or the W3C version |
| 16:24 | <Dashiva> | I wouldn't link my enemies to the single-page version, much less an innocent stranger ;) |
| 16:25 | <miketaylr> | heh. no need for a link, i've been there plenty. ;) |
| 16:26 | <miketaylr> | ok found it in 4.10.16.3 The constraint validation API, thanks again annevk2 |
| 16:29 | <annevk2> | anytime |
| 16:55 | <zcorpan_> | let's see if that's controversial... |
| 16:56 | <jgraham> | zcorpan_: +1 |
| 17:27 | <ROBOd> | i am another user of the spec who is missing http://www.whatwg.org/specs/web-apps/current-work/multipage/ |
| 17:27 | <ROBOd> | could someone fix that? |
| 17:27 | <annevk2> | come back tomorrow |
| 17:27 | <Philip`> | I would fix it if my computer that generated the multipage spec wasn't suffering from a lack of internets |
| 19:49 | <zcorpan_> | hmm, http://jsbeautifier.org/ has <script>...<!--...</script> |
| 19:49 | <zcorpan_> | and doesn't use quirks mode |
| 19:53 | <zcorpan_> | it would break with my idea of handling it, because one </script> gets eaten, causing the script to not compile and the next script to be ignored, too |
| 19:54 | <zcorpan_> | it would work with hsivonen's idea though, because the <!-- is not only preceded by whitespace |
| 20:01 | <zcorpan_> | oh actually |
| 20:01 | <zcorpan_> | it doesn't fail |
| 20:01 | <zcorpan_> | yay |
| 20:02 | <zcorpan_> | because it doesn't have "<script" after the <!-- |
| 20:06 | zcorpan_ | writes down a slightly different idea in http://wiki.whatwg.org/wiki/CDATA_Escapes#Proposal_.232 |
| 20:07 | <Philip`> | Proposal 232? Must be a pretty hard problem |
| 20:08 | <zcorpan_> | proposal .232 |
| 20:09 | <zcorpan_> | Philip`: any idea on how to test the different ideas againt content? |
| 20:10 | <zcorpan_> | i guess one thing that's simple to test is whether there are pages with the end tag in an escaped text span for (r)cdata elements other than script |
| 20:11 | <zcorpan_> | whether they need the escapedness or would be better off without it |
| 20:11 | <Philip`> | Get someone to implement them all, run them all against lots of pages, find which pages produce different DOMs with the different methods, and then manually check what seems to be the best behaviour in each case? |
| 20:11 | <Philip`> | (assuming there is a finite list of proposals) |
| 20:22 | <jgraham> | You should really test against reparsing since that is the current behaviour |
| 20:26 | <jgraham> | (so proposals that fail to match reparsing in a significant number of cases are a no-go) |
| 20:28 | <zcorpan_> | so who feels like implementing reparsing and proposal #2 and #3 in html5lib or v.nu? |
| 20:28 | <zcorpan_> | maybe there's a simpler way to find out if they are workable |
| 20:29 | <Philip`> | If someone implements it in html5lib, don't expect me to try running it on half a million pages :-p |
| 20:30 | <zcorpan_> | because html5lib is too slow? |
| 20:30 | <Philip`> | Yes |
| 20:30 | <zcorpan_> | ok |
| 20:30 | <Philip`> | and I don't fancy running it for days |
| 20:32 | <zcorpan_> | don't we have some data about <!-- in (r)cdata elements somewhere already? |
| 20:33 | <zcorpan_> | i mean list of pages |
| 20:37 | <zcorpan_> | http://philip.html5.org/data/pages-with-unclosed-comments.txt |
| 20:38 | <Philip`> | http://philip.html5.org/data/pages-with-unclosed-scripts-and-comment-stuff.txt |
| 20:38 | <Philip`> | Please don't ask me what that regexp does |
| 20:39 | <zcorpan_> | quite a few would break with proposal #2 |
| 20:39 | <zcorpan_> | because they use <!-- as normal but have no --> at all |
| 20:40 | <zcorpan_> | but my idea seems to work |
| 20:41 | <Dashiva> | Philip`: It doesn't look that bad |
| 20:42 | <TabAtkins> | I find it odd that sites creating today still use the <!-- hack on their <script>s at all. |
| 20:42 | <TabAtkins> | We're *long* past the point where that was necessary. |
| 20:44 | <gsnedders> | Philip`: What does that regexp do? |
| 20:44 | <Philip`> | /ignore gsnedders |
| 20:44 | <Philip`> | Ask Dashiva what it does |
| 20:44 | <gsnedders> | Oh, I can see myself. |
| 20:44 | <Dashiva> | It finds <script followed by <!-- followed by </script> before any occurence of -->, and the ensures there's no later --> </script> |
| 20:45 | <gsnedders> | I just asked you because you said not to do so. |
| 20:45 | <gsnedders> | And I'm a bitch. |
| 20:46 | <zcorpan_> | ok so that leaves proposal #3 |
| 20:46 | <zcorpan_> | go find a page that breaks with proposal #3 |
| 20:49 | <gsnedders> | … Simon says, ordering his minions around :P |
| 20:49 | <Philip`> | Dashiva: I don't think that's right |
| 20:50 | <Philip`> | It ensures there *is* a later -->, but no later </script> |
| 20:51 | <Dashiva> | That's what I meant, yes |
| 20:51 | <Philip`> | so the comment has to be closed, but the <script> will not be closed because the only </script> was inside the comment |
| 20:52 | <zcorpan_> | that's still not right |
| 20:52 | <zcorpan_> | if there is a later -->, it ensures that there's no later </script> |
| 20:53 | <zcorpan_> | but there can be a later </script> without --> before |
| 20:53 | <zcorpan_> | or no --> at all |
| 20:57 | <Philip`> | It's like <script> <!-- {anything except -->} </script> {anything except </script>} --> {anything except </script>} {end of file} |
| 20:57 | <Philip`> | I think |
| 20:58 | <Dashiva> | The --> can be repeated |
| 20:58 | <Philip`> | Only because --> is anything except </script> |
| 20:58 | <Dashiva> | Maybe should've been ? instead of * at the end for clarity |
| 20:59 | <Philip`> | The * is just implementing the {anything except </script>} |
| 20:59 | <Philip`> | because it's zero-or-more characters that are not '<', or are '<' but not followed by '/script' |
| 20:59 | <Dashiva> | Oh, right |
| 20:59 | <Dashiva> | I misread the nesting |
| 21:05 | <zcorpan_> | it works for the mozilla bugs hsivonen referenced |
| 21:05 | <zcorpan_> | and it works for the pages in Philip`'s list |
| 21:06 | <zcorpan_> | and it works for <script><!-- document.write('<script></script>'); document.write('<script></script>'); </script> |
| 21:06 | <zcorpan_> | it does not work for <script><!-- document.write('<sc'+'ript></script>'); </script> |
| 21:07 | gsnedders | wonders where to eat on Sunday evening |
| 21:07 | <zcorpan_> | doesn't work for <script><!-- document.write('<sc'+'ript></script>'); --></script> either |
| 21:08 | <zcorpan_> | i wonder if people split the start tag but not the end tag in their d.write |
| 21:08 | <jgraham> | gsnedders: Which country will you be in? |
| 21:08 | <gsnedders> | jgraham: Sweden |
| 21:09 | <jgraham> | gsnedders: Why are you wondering? |
| 21:09 | <gsnedders> | jgraham: I can't get any of my stuff from the office until Monday, so I don't have a plate or anything useful… |
| 21:09 | <jgraham> | gsnedders: We can provide you with food, probaby |
| 21:10 | <gsnedders> | The second time in a month I've scrounged food off you in two months :) |
| 21:10 | <gsnedders> | s/two months/one month/ |
| 21:10 | <gsnedders> | Huh… |
| 21:10 | <gsnedders> | That still makes no sense. |
| 21:10 | <jgraham> | indeed. But since the first attemp made my head explode |
| 21:10 | <gsnedders> | s/in one month/on a Sunday evening/ |
| 21:10 | <jgraham> | there's no point in correcting it now |
| 21:11 | <gsnedders> | Sorry. I guess I won't get food if you're headless now. |
| 21:11 | <jgraham> | But yeah, let me know when you want to eat or something. |
| 21:11 | <jgraham> | (I guess you arrive on Sunday then?) |
| 21:11 | <gsnedders> | jgraham: I guess 7ish or later would suit me |
| 21:11 | <Dashiva> | Doesn't not splitting the end tag break? |
| 21:11 | <gsnedders> | jgraham: Yeah, I arrive 16:50 from Skavsta |
| 21:12 | <jgraham> | gsnedders: Sounds fine |
| 21:12 | <gsnedders> | jgraham: Should I just come over at 7ish then? I can call if anything happens which means that won't work. |
| 21:13 | <zcorpan_> | Dashiva: it breaks with my proposal but doesn't break today |
| 21:14 | gsnedders | goes back to the book that shall not be named |
| 21:16 | zcorpan_ | writes down another potential thing that would break with his proposal |
| 21:21 | <zcorpan_> | hmm |
| 21:22 | <zcorpan_> | <script><!-- document.write('<scr'+'ipt></script>'); //--></script> might not be too uncommon on the face of it |
| 21:30 | <zcorpan_> | (?is)<script>\s*<\!--(([^-]|-(?!->)|([^<]|<(?!script[\s\/>]))*<\/script[\s\/>] |
| 21:30 | <zcorpan_> | is that correct? |
| 21:30 | <jgraham> | gsnedders: Yes |
| 21:31 | <zcorpan_> | Philip`, Dashiva ^ |
| 21:32 | <zcorpan_> | basically i want to find pages that have </script inside escaped text span but without <script before it |
| 21:49 | <cying> | Hixie: looks like the html5 spec is missing: http://www.whatwg.org/specs/web-apps/current-work/multipage/ |
| 21:50 | <Philip`> | cying: Known problem |
| 21:50 | <cying> | Philip`: ah thanks |
| 21:50 | <Philip`> | zcorpan_: It's not correct |
| 21:51 | <zcorpan_> | snap |
| 21:51 | <Philip`> | zcorpan_: e.g. you've effectively got [^-]|[^<]|... and every character is going to match that |
| 21:52 | gsnedders | thinks we should follow the OED's example and produce a "Compact Edition" in 4pt type |
| 21:53 | <Philip`> | Oh, good, my computer's back |
| 21:53 | Philip` | regenerates the spec |
| 21:53 | <Philip`> | cying: Should work now |
| 21:54 | <zcorpan_> | Philip`: oops |
| 21:54 | <zcorpan_> | Philip`: how about... |
| 21:54 | <zcorpan_> | (?is)<script>\s*<\!--(([^-]|-(?!->))|([^<]|<(?!script[\s\/>])))*<\/script[\s\/>] |
| 21:57 | <Philip`> | That seems to be effectively identical to the previous version |
| 21:58 | <zcorpan_> | really? |
| 21:58 | <Philip`> | because (x|y|z) and ((x|y)|z) are equivalent |
| 21:58 | <Philip`> | which is the only change I see |
| 21:59 | <zcorpan_> | so how do i write any sequence of characters except --> or <script followed by whitespace, slash or greater-than? |
| 22:00 | <Philip`> | I think you may want something more like ([^<-]|-(?!->)|<(?!script[\s/>])) |
| 22:00 | <zcorpan_> | ah |
| 22:00 | <zcorpan_> | ok |
| 22:01 | <Philip`> | ([^x]|[^y] is not equivalent to [^xy], and you want the latter) |
| 22:01 | <zcorpan_> | yeah, makes sense |
| 22:01 | <Philip`> | Also you probably want to allow attributes in the first <script> tag |
| 22:01 | <zcorpan_> | oops |
| 22:01 | <Philip`> | Also you don't need to escape ! |
| 22:02 | <Philip`> | Also you don't need the (?s) (it just affects the definitions of ^ and $) |
| 22:02 | <Philip`> | Also you don't need to escape / |
| 22:02 | <Philip`> | (...unless you're using it in a language with / as the regexp delimiter) |
| 22:03 | <Philip`> | (...but my grep stuff doesn't use delimited regexps, so it doesn't need to be escaped) |
| 22:03 | <Philip`> | Otherwise it looks fine :-) |
| 22:04 | <Philip`> | though I could be mistaken |
| 22:05 | <zcorpan_> | (?i)<script[\s/]?[^>]*>\s*<!--([^<-]|-(?!->)|<(?!script[\s/>]))*</script[\s/>] |
| 22:06 | <Philip`> | The [\s/]? is redundant with the [^>]* |
| 22:07 | <zcorpan_> | yeah |
| 22:07 | <zcorpan_> | could you try the regexp on a set of pages? :) |
| 22:09 | <Philip`> | (?i)<script[^>]*>\s*<!--([^<-]|-(?!->)|<(?!script[\s/>]))*</script[\s/>] |
| 22:09 | <Philip`> | Do you just want a list of pages, or also the matching substrings? |
| 22:10 | <zcorpan_> | matching substrings would be nice |
| 22:11 | <zcorpan_> | but not essential |
| 22:13 | <Philip`> | zcorpan_: Something like http://philip.html5.org/data/script-close-in-escape-without-script-open.txt ? |
| 22:13 | <Philip`> | (That's not the whole output, just the first small fraction) |
| 22:14 | <Philip`> | The formatting is pretty rubbish, but I don't know if that's okay |
| 22:14 | Philip` | can change it fairly easily |
| 22:15 | <zcorpan_> | it's ok |
| 22:18 | <zcorpan_> | hmm actually, the \s* doesn't have to be whitespace |
| 22:21 | <zcorpan_> | (?i)<script[^>]*>([^<]|<(?!/script[\s/>])*<!--([^<-]|-(?!->)|<(?!script[\s/>]))*</script[\s/>] |
| 22:21 | <zcorpan_> | could you rerun it with that instead? |
| 22:21 | <Philip`> | Hmm, odd, it seems to have finished already |
| 22:21 | <zcorpan_> | ah |
| 22:21 | <Philip`> | which is surprisingly quick |
| 22:23 | <Philip`> | zcorpan_: http://philip.html5.org/data/script-close-in-escape-without-script-open.txt is the old one |
| 22:23 | <Philip`> | Running again with the new one... |
| 22:23 | <Philip`> | Oh, syntax error |
| 22:23 | <Philip`> | I think you missed a ) |
| 22:24 | <zcorpan_> | oops, yeah |
| 22:24 | <zcorpan_> | (?i)<script[^>]*>([^<]|<(?!/script[\s/>]))*<!--([^<-]|-(?!->)|<(?!script[\s/>]))*</script[\s/>] |
| 22:26 | <Philip`> | zcorpan_: Remind me to check this in five minutes |
| 22:26 | <zcorpan_> | sure |
| 22:27 | <zcorpan_> | the current output shows lots of examples that breaks with proposal #2 |
| 22:29 | <zcorpan_> | did the google analytics boilerplate lack the --> at some point? |
| 22:31 | <zcorpan_> | www.grandparents.com/gp/content/expert-advice/family-matters/article/thatevildaughterinlaw.html has unescaped end tag in d.write :( |
| 22:31 | <zcorpan_> | document.write('<SCR' + 'IPT LANGUAGE=JavaScript1.1 SRC="' + OAS_url + 'adstream_mjx.ads/' + OAS_sitepage + '/1' + OAS_rns + '@' + OAS_listpos + '?' + OAS_query + '"></SCRIPT> |
| 22:38 | <zcorpan_> | Philip`: reminder |
| 22:39 | <Philip`> | http://philip.html5.org/data/script-close-in-escape-without-script-open-2.txt |
| 22:40 | <zcorpan_> | thanks! |
| 22:42 | <zcorpan_> | <script type="text/javascript">--><!-- |
| 22:42 | <zcorpan_> | amazon_ad_tag = "go2urlcomtheu-20"; ... amazon_ad_include = "marketing";//--> |
| 22:42 | <zcorpan_> | <!-- </script> |
| 22:42 | <zcorpan_> | wtf? |
| 22:43 | <zcorpan_> | ok, that breaks with my proposal, too |
| 22:43 | <zcorpan_> | maybe it needs to be combined with the variant of proposal #2 |
| 22:44 | <zcorpan_> | so that only <script>\s*<!-- starts escaped text span |
| 22:44 | Philip` | supposes some amount of breakage may be inevitable |
| 22:44 | <zcorpan_> | yeah |
| 22:46 | <zcorpan_> | oh wait, it doesn't break with my proposal |
| 22:49 | <zcorpan_> | wow having <!--</script> is actually pretty common in this set |
| 22:49 | <zcorpan_> | (or actually //--><!--</script>) |
| 22:51 | <zcorpan_> | www.celebrity-link.com/c106/showcelebrity_categoryid-10687.html breaks |
| 22:56 | <zcorpan_> | www2.solomoto.es/enviar-noticia.asp?noti=44557&pag=noticia has the same snippet as grandparents.com |
| 22:58 | <zcorpan_> | me.yaplog.jp/viewBoard.blog?boardId=975 breaks too |
| 23:09 | <zcorpan_> | maybe the parser should look at "document.write" instead |
| 23:10 | <zcorpan_> | and ignore "</script" until the next ")" |
| 23:20 | <zcorpan_> | hmm, i guess that would break pages that put the string in a variable and then do document.write(str) |
| 23:21 | <Philip`> | And pages that do var w=document.write; w("str") |
| 23:22 | <Philip`> | and it'd be an incredibly gross hack |
| 23:33 | <zcorpan_> | ok time to sleep |
| 23:33 | <zcorpan_> | nn |
| 23:49 | <hober> | http://lists.w3.org/Archives/Public/public-html/2009Sep/att-1216/MicrosoftDistributedExtensibilitySubmission.htm |
| 23:52 | <othermaciej> | time for someone to dig up Opera's old data |
| 23:58 | <Philip`> | "The proposal as stated closely matches behavior that Internet Explorer has had for a number of releases" |
| 23:58 | <Philip`> | That seems untrue |
| 23:58 | <Philip`> | because e.g. IE has never had proper Namespace-like DOM attributes |