| 00:05 | <annevk> | oh man |
| 00:05 | <annevk> | http://daringfireball.net/linked/2010/06/01/thurrott-understanding is brilliant |
| 00:06 | <annevk> | I really dig the whole quote what the expert said before the fact after the fact thing Gruber has going on |
| 00:09 | <annevk> | "First Draft of SPARQL 1.1 Federation Extensions Published; Five SPARQL 1.1 Drafts Updated" federation extensions? |
| 00:09 | <annevk> | gotta admit that this sounds pretty serious |
| 00:09 | <annevk> | serious and boring |
| 01:17 | <mikl0> | http://pastebin.com/iLAHWMAR can someone explain this DOM behavior to me? |
| 01:20 | <AryehGregor> | mikl0, browsers do not support the self-closing /> syntax in text/html. |
| 01:20 | <AryehGregor> | You have to use an explicit closing tag like </firstChild>. |
| 01:21 | <mikl0> | so i must use <firstChild></firstChild>. however if i do XHTML it will work as that does support self closing right? |
| 01:22 | <Philip`> | mikl0: Yes, if by "do XHTML" you mean serving the file as application/xhtml+xml |
| 01:22 | <Philip`> | (not just changing the doctype or the <meta>) |
| 01:22 | <mikl0> | AryehGregor: thanks for always asnwering i learn so much. |
| 01:23 | <mikl0> | Philip`: yes doctype, meta, http header, and .xhtml extension. thanks too Philip` |
| 01:23 | <tabatkins> | Dammit, my computer's broken. |
| 01:23 | <Philip`> | TabAtkins: Buy a new one |
| 01:24 | <TabAtkins> | "computer" is defined very loosely here. Something between me and the rest of the internet is broken, and causing all of my requests to take a retarded amount of time. |
| 01:25 | <TabAtkins> | And... now it's back. |
| 01:25 | <AryehGregor> | TabAtkins, make sure no one on your network is using P2P or other bandwidth-heavy stuff. |
| 01:25 | <AryehGregor> | I saw 600 ms ping recently to my ISP when another computer in the house was making an initial backup to Mozy. |
| 01:26 | <TabAtkins> | I'm on Google's network, so I have no idea how I would check, nor what I would do if they were. |
| 01:26 | <AryehGregor> | Oh, right. |
| 01:26 | <AryehGregor> | I thought you were at home. |
| 01:26 | <Philip`> | I would have thought Google would have enough bandwidth to cope with someone using P2P software |
| 01:27 | <TabAtkins> | AryehGregor: You're 3 hours later than me. I'm at work for another half hour. |
| 01:27 | <TabAtkins> | Philip`: One would think so, yes. I suspect it was something crazy related to startup stuff, since I just restarted my computer. |
| 01:27 | <AryehGregor> | Time is meaningless to me. I've gotten up and have not yet gone to bed, that's all I notice. |
| 01:28 | <AryehGregor> | That's a slight exaggeration, but things like "working hours" are certainly meaningless to me. |
| 01:28 | <TabAtkins> | Somehow I was using 6 gigs of RAM between a handful of browser windows doing non-intensive things, and a few idle consoles. |
| 01:28 | <AryehGregor> | o_O |
| 01:28 | <AryehGregor> | Not counting buffers and cache? |
| 01:28 | <gsnedders> | TabAtkins: Have you been looking at memory usage bugs? |
| 01:29 | <TabAtkins> | gsnedders: Not on purpose. ^_^ |
| 01:29 | <TabAtkins> | Anyway, 20 days of uptime, so shrug. Same thing happened after roughly 20 days of uptime previously. |
| 01:29 | <AryehGregor> | What OS? |
| 01:29 | <TabAtkins> | GUbuntu. |
| 01:30 | <AryehGregor> | I had XOrg using 1.5G of memory a while back, but that was after more like 90d of uptime. |
| 01:30 | <gsnedders> | Yeah, I've had that happen too |
| 01:34 | <AryehGregor> | This is why non-memory-managed languages stink. But I guess we have to live with them. |
| 01:35 | <gsnedders> | memory-managed languages stink because their GC tends to have bugs too :) |
| 01:39 | <TabAtkins> | Single source of bugs in an area where the bugs are relevant to the very operation of the component is better than bugs distributed throughout every program that aren't critical to the operation of the programs. |
| 01:40 | <AryehGregor> | Well, I mostly write in PHP. Despite the PHP developers' most strenuous efforts to introduce bugs and outright insanity into all parts of the reference language implementation, it's hard to get memory to leak past the end of a request. |
| 01:40 | <Philip`> | Memory-managed languages don't protect you from poor caching strategies, which is sometimes what causes painful memory usage |
| 01:40 | <AryehGregor> | Well, it's an improvement. |
| 01:47 | <Philip`> | AryehGregor: It's better than nothing, there's just a danger that people think GC means they no longer have to worry about the lifetimes of allocated objects |
| 01:48 | <AryehGregor> | From experience, it seems like people don't even notice the GC once they're used to it. It doesn't make them less able to troubleshoot memory usage. |
| 01:48 | <Philip`> | when really you still ought to worry and the GC just means you don't have any explicit end-of-lifetime markers in your code |
| 01:48 | <AryehGregor> | It means that you almost never have to worry about explicitly deallocating things, not never. |
| 01:48 | <AryehGregor> | I've cut memory usage using unset() in PHP before. |
| 01:50 | Philip` | supposes he's just mildly unhappy with how Eclipse seems to be using 1GB of memory, when he's simply been using it as a fancy text editor for a day |
| 02:04 | AryehGregor | has never noticed vim using an appreciable amount of memory. |
| 07:06 | <MikeSmith> | hsivonen: I have a question about if/how to use the v.nu HTML5 parser in conjunction with existing command-line Java apps |
| 07:06 | <MikeSmith> | background is this - |
| 07:06 | <MikeSmith> | if I want to use a different parser with, e.g., Saxon |
| 07:07 | <MikeSmith> | instead of the built-in parser that Saxon uses |
| 07:07 | <MikeSmith> | Aelfred I guess |
| 07:07 | <MikeSmith> | then I can do this: |
| 07:07 | <MikeSmith> | java \ |
| 07:07 | <MikeSmith> | -Djavax.xml.parsers.DocumentBuilderFactory=\ |
| 07:07 | <MikeSmith> | org.apache.xerces.jaxp.DocumentBuilderFactoryImpl \ |
| 07:07 | <MikeSmith> | -Djavax.xml.parsers.SAXParserFactory=\ |
| 07:07 | <MikeSmith> | org.apache.xerces.jaxp.SAXParserFactoryImpl \ |
| 07:07 | <MikeSmith> | com.icl.saxon.StyleSheet \ |
| 07:07 | <MikeSmith> | ... |
| 07:08 | <MikeSmith> | and that'll cause Saxon to use Xerces instead |
| 07:08 | <MikeSmith> | as far as I can tell, I can't currently do that with the v.nu HTML parser, right? |
| 07:09 | <MikeSmith> | so is there some other way I could cause Saxon to use the v.nu parser instead? |
| 07:09 | <MikeSmith> | or could we add support for those properties to the v.nu parser? |
| 07:09 | <MikeSmith> | that's some JAXP thing, right? |
| 07:14 | MikeSmith | takes a look at the XSLT4HTML code for the time being |
| 07:27 | <fantasai> | What's the default output encoding for html5lib, utf-8? |
| 07:28 | <MikeSmith> | fantasai: yeah |
| 07:28 | <fantasai> | MikeSmith: or is it the system default encoding? |
| 07:29 | <MikeSmith> | I believe it is utf-8 |
| 07:29 | <MikeSmith> | not whatever the default system encoding is |
| 07:32 | MikeSmith | looks at some code |
| 07:33 | <MikeSmith> | hmm, I find "serializer.HTMLSerializer(**kwargs).serialize(tokens, encoding='utf-8')" in code I'm using |
| 07:34 | <MikeSmith> | so maybe it's not actually utf-8 by default |
| 07:34 | MikeSmith | glances around for help |
| 07:38 | <MikeSmith> | checking the html5lib source, it seems that html5lib/serializer code does not itself set encoding to utf-8 |
| 07:39 | <MikeSmith> | fantasai: so I think I was wrong |
| 07:39 | <MikeSmith> | I can't see anywhere in the html5lib code itself where the output encoding is set to utf-8 |
| 08:07 | <Mrmil> | Anyone has an idea why wrap="off" for textarea is not present in the spec? |
| 08:08 | <zcorpan_> | Mrmil: it's semantically equivalent to wrap="soft" |
| 08:09 | <annevk> | rendering is defined |
| 08:19 | <Mrmil> | wrap soft/hard doesn't seem to work (or I'm not getting it right), maybe browsers don't support it? |
| 09:10 | <MikeSmith> | listening to introductory presentation related to container-based virtualization |
| 09:10 | <MikeSmith> | network namespace isolation |
| 09:12 | <kennyluck> | It looks like a hardware-related talk to me, so I wasn't interested. |
| 09:16 | <MikeSmith> | kennyluck: it's not about software |
| 09:17 | <kennyluck> | OK |
| 09:17 | <MikeSmith> | kernel, socket and routing library code |
| 09:17 | <MikeSmith> | dude is from Tsinghua university |
| 09:18 | <kennyluck> | Ah, it still sounds interesting. |
| 09:18 | <kennyluck> | Yeah, I noticed that, FYI ,there's a Tsuinghua uni. in Taiwan and a Tsuinghua uni. in Mainland China. |
| 09:18 | <kennyluck> | I was wondering which one does he come from. |
| 09:19 | <MikeSmith> | solution slide: "associate each uid with an IP address" |
| 09:19 | <MikeSmith> | kennyluck: from accent, I guess mainland |
| 09:19 | <kennyluck> | Oh, that sounds quite crazy. |
| 09:20 | <kennyluck> | ref. "associate each uid with an IP address" |
| 09:20 | <MikeSmith> | http://www.nrc.tsinghua.edu.cn/7_english/profile/chenmaoke.htm |
| 09:32 | <jgraham> | So the Decisions seemed to mainly take into account what people said on the poll even though it said not to say anything in the poll unless it wasw not already in a change proposal |
| 09:32 | <jgraham> | Oh, TabAtkins said the same thing on the list |
| 09:35 | <jgraham> | and if I understand what Sam said it seems like the weight given to change proposals is asymmetric so only arguments in favour of change are considered |
| 09:35 | jgraham | is confused |
| 09:37 | <hsivonen> | jgraham: yeah, I think the instructions given by the chairs pre-poll and the explanations post-poll don't go well together |
| 09:38 | <hsivonen> | frankly, I'm not at all surprised |
| 09:41 | <hsivonen> | TabAtkins: maybe next time in addition to submitting a no-change Change Proposal, you should also object to the deletionist Change Proposal by saying that something like "I repeat and raise as objections against this Proposal all the points in favor of no change raised in the no-change Proposal as if fully restated herein" |
| 09:43 | <hsivonen> | jgraham: which is why I took the time to file an objection that was slightly different from the points made in the no-change Proposal |
| 09:46 | <jgraham> | hsivonen: It appears to have been worthwhile. But should have been unnecessary |
| 09:48 | <hsivonen> | In retrospect, I probably should have copied and pasted my objection to other polls too |
| 09:48 | <hsivonen> | just in case |
| 10:16 | <Hixie> | hsivonen, jgraham: you could decode sam's e-mails and figure out what the decisions were? I wasn't even able to penetrate his wishy washy language to work out what to change |
| 10:17 | <Hixie> | btw hsivonen you might want to comment on https://bugs.webkit.org/show_bug.cgi?id=40047 |
| 10:34 | <daedb_> | Oh yay, there's fun stuff on public-html again. Nice conspiracy theories :D |
| 10:36 | <Hixie> | is there some equivalent of svn blame that will tell me what revisions subsequently _removed_ lines that were checked in at a particular revision? |
| 10:46 | <annevk> | Hixie, scan the commit messages? :) |
| 10:47 | <Hixie> | there are 1000s :-) |
| 10:48 | <Philip`> | grep and a bisect tool? |
| 10:49 | <Philip`> | like http://search.cpan.org/dist/App-SVN-Bisect/bin/svn-bisect |
| 10:49 | <annevk> | you can do an search through them |
| 10:50 | <Hixie> | so "no"? :-) |
| 10:50 | <jgraham> | So if I have <iframe src="javascript:something"> in the source of a document, what happens? How many load events do I get? |
| 10:51 | <jgraham> | (like is there an implicit load of about:blank first?) |
| 10:59 | <zcorpan_> | jgraham: seems browsers differ |
| 10:59 | <zcorpan_> | javascript:void() gives 1 load in opera and chrome, 0 in firefox |
| 10:59 | <zcorpan_> | javascript:'' gives 1 in opera and firefox, 2 in chrome |
| 11:00 | <jgraham> | zcorpan_: Any idea what the spec says, or whether it is defensible? |
| 11:00 | <zcorpan_> | haven't looked at the spec |
| 12:15 | <hsivonen> | Hixie: regarding https://bugs.webkit.org/show_bug.cgi?id=40047 , I think Gecko may still be blocking on inline scripts if there are pending sheets |
| 12:16 | <RiPPER> | hi |
| 12:27 | <hsivonen> | Hixie: actually, I think the spec in now racy in the blocking behavior of inline scripts |
| 12:27 | <hsivonen> | sigh |
| 12:28 | <Dashiva> | The smell of Process in the morning... |
| 12:28 | <RiPPER> | I read somewhere that in html 5 I can use xhtml syntax, is this right? |
| 12:29 | <Dashiva> | No, but some xhtml-isms are conforming |
| 12:29 | <Dashiva> | Like <br/> instead of <br> |
| 12:29 | <hsivonen> | RiPPER: you can 1) use an XML serialization of HTML5 when serving content as application/xhtml+xml or you can 2) use certain XHTMLisms in text/html |
| 12:30 | <hsivonen> | RiPPER: you can't use the whole of XHTML syntax in text/html, though |
| 12:30 | <Hixie> | hsivonen: really? i thought we'd fixed all the racy behaviour... |
| 12:32 | <hsivonen> | Hixie: by inspection, Gecko's implementation sure looks racy... |
| 12:33 | <Hixie> | hsivonen: well if the spec is racy, file a bug |
| 12:35 | <hsivonen> | Hixie: the spec isn't racy, it seems. But the spec looks unacceptable for implementation |
| 12:36 | <Hixie> | well, file a bug for that too |
| 12:36 | <hsivonen> | both to me and, it seems, to Eric Seidel per comment #2 in the WebKit bug |
| 12:37 | <hsivonen> | since it would be pretty crazy to spin a nested event loop from within document.write() in order to wait for a style sheet |
| 12:37 | <hsivonen> | I'll file a spec bug |
| 12:37 | <Hixie> | well how else are you going to wait for a style sheet? |
| 12:38 | <hsivonen> | Hixie: do we know that document.written inline scripts need to wait for sheet for Web compat? |
| 12:38 | <Hixie> | iirc that's what browsers did when i tested it |
| 12:39 | <Hixie> | but who knows |
| 12:39 | <hsivonen> | by inspection, it seems to me that Gecko doesn't do that |
| 12:39 | <annevk> | Gecko used to have synchronous style sheet loading |
| 12:39 | <Hixie> | making things dependent on whether they'd inserted via document.write() or not seems pretty wacky |
| 12:40 | <hsivonen> | my axiom is that nested event loops are wacky |
| 12:40 | <Hixie> | well on a side note, technically the html5 spec doesn't have any nested event loops |
| 12:40 | <Hixie> | but that's neither here nor there |
| 12:41 | <Hixie> | my point is that since we need to spin the event loop for regular <script>s, that boat has sailed |
| 12:41 | <hsivonen> | that doesn't help much, since we are dealing with C++ method call semantics |
| 12:41 | <hsivonen> | so there aren't fancy yields and continuations available |
| 12:41 | <Hixie> | so why make an exception for document.written ones? |
| 12:41 | <annevk> | I think we had the ability for some time to stop script execution when a layout DOM attribute was asked for to be able to fetch the necessary style sheets and compute something... Mostly to work with sites that were depending on the synchronous style sheet loading. I suspect things have changed though. For one Gecko no longer does the synchronous loading... |
| 12:42 | <hsivonen> | Hixie: because for regular scripts you don't need a nested event loop in order to be able to wait for the sheet |
| 12:42 | <Hixie> | without the event loop, the sheet can never load |
| 12:42 | <Hixie> | since the load happens on the event loop |
| 12:42 | <hsivonen> | Hixie: when you tested Gecko, did you make sure to document.write the <link rel=stylesheet> and the inline script together? |
| 12:42 | <Hixie> | no idea |
| 12:43 | <hsivonen> | if the <link rel=stylesheet> is in the network stream, you block whatever script would call document.write before you get to the document.write |
| 12:43 | <jgraham> | hsivonen: (you can do fancy yield and continuation in C++ if you design for it, of course) |
| 12:43 | <jgraham> | (I'm not suggesting that doesn't make it too high an implementation burden to require it) |
| 12:44 | <hsivonen> | jgraham: do you mean with arbitrary methods on the stack and without breaking the abstraction that the program isn't supposed to mess with its own stack as raw memory? |
| 12:45 | <Hixie> | you can do things with setjmp/longjmp |
| 12:45 | <Hixie> | but i wouldn't recommend it |
| 12:45 | <Hixie> | you can also simply write your code so that the continuation is a separate method |
| 12:46 | <Hixie> | it's not like the spinning is in an arbitrary place, it's in a well-defined place |
| 12:46 | <Hixie> | but in practice it doesn't matter much |
| 12:46 | <jgraham> | hsivonen: I am not familiar with the range of implementation approaches, I just know it is possible |
| 12:46 | <Hixie> | i was just saying the spec doesn't technically use nested event loops |
| 12:47 | <hsivonen> | Hixie: that's not much of a consolation |
| 12:47 | <Hixie> | i don't think anybody intended it to be a consolation :-) |
| 13:59 | <jgraham> | Hmm, is it generally true that the list of possibilities in the tree construction algorithm can be treated as unordered? It is at least untrue in the foreignContent case, but it seems like it would be nice if it was always true |
| 14:00 | <jgraham> | (apart from the "anything else" type clauses of course) |
| 14:06 | <hsivonen> | jgraham: IIRC, it's unordered |
| 14:07 | <jgraham> | Hmm, what am I missing about EOF handling in foreign content? It seems like it doesn't switch insertion modes but reprocesses the token |
| 14:07 | <jgraham> | So it will just pop everything off the stack of open elements |
| 14:09 | <jgraham> | hsivonen: It isn't in the case of 'A start tag whose tag name is one of: "b", "big", [...]' because that rule only applies if you don't match the rule 'A start tag, if the current node is an element in the HTML namespace.' |
| 14:10 | <hsivonen> | jgraham: oh ok. |
| 14:11 | <jgraham> | hsivonen: Any clues about what I am missing with the EOF thing? |
| 14:17 | <zcorpan_> | so what happened to binary arrays? |
| 14:17 | <hsivonen> | jgraham: I think you found a spec bug. |
| 14:18 | <hsivonen> | jgraham: before reprocessing the token, the tree builder should switch to the original mode |
| 14:18 | <hsivonen> | or at least that's what I have implemented |
| 14:19 | <jgraham> | hsivonen: For everything in that clause or just the EOF case? |
| 14:34 | <jgraham> | hsivonen: Filed a bug |
| 14:35 | <hsivonen> | jgraham: I meant the EOF case, but there may be other bugs |
| 14:35 | <jgraham> | I think the other case works either way |
| 14:36 | <jgraham> | But I may be wrong |
| 14:51 | <mikekelly> | hi fans |
| 14:52 | <mikekelly> | I've come for my bi-annual discussion about the type attribute |
| 14:53 | <Dashiva> | Please make sure you have new information before reopening old issues |
| 14:54 | <mikekelly> | I'm joking |
| 14:56 | <mikekelly> | why is it not possible for javascript to kick off a page transition and have control over the headers in the same as you do with XHR ? |
| 14:57 | <annevk> | so you were not joking? |
| 14:58 | <annevk> | yawn |
| 14:58 | <m_ali> | Hi |
| 14:59 | <m_ali> | i have a project that is using html5lib for sanitization of user input |
| 15:00 | <m_ali> | it is using the tokenizer and code like this: |
| 15:00 | <m_ali> | p = html5lib.HTMLParser(tokenizer=HTMLSanitizer, |
| 15:00 | <m_ali> | tree=treebuilders.getTreeBuilder("dom")) |
| 15:00 | <m_ali> | dom_tree = p.parseFragment(html) |
| 15:00 | <m_ali> | walker = treewalkers.getTreeWalker("dom") |
| 15:00 | <m_ali> | stream = walker(dom_tree) |
| 15:00 | <m_ali> | s = serializer.HTMLSerializer(omit_optional_tags=False, |
| 15:00 | <m_ali> | quote_attr_values=True) |
| 15:00 | <m_ali> | output_generator = s.serialize(stream) |
| 15:00 | <m_ali> | #logging.debug(''.join(output_generator)) |
| 15:00 | <m_ali> | return u''.join(output_generator) |
| 15:00 | <m_ali> | I am looking to allow the user to embed videos which are in <object> tags and the sanitization changes that to text. |
| 15:00 | <m_ali> | Any ideas what I need to do? |
| 15:01 | <zcorpan_> | if you allow <object>, there's no point in sanitizing at all |
| 15:02 | <mikekelly> | annevk: I was joking.. do you know the answer to that question? |
| 15:03 | <zcorpan_> | unless i guess you require data='' to be something that is trusted |
| 15:03 | <m_ali> | I can check the object tag that contains the allowed video sites |
| 15:06 | <TabAtkins_> | dammit, lachy. why you gotta beat me to posting things? |
| 15:07 | <m_ali> | @zcorpan_: would I need to look at each token in the tokenizer or is there any other way? |
| 15:08 | <jgraham> | m_ali: YOu can allow object in the sanitizer. Restricting the allowed attribute values is harder |
| 15:09 | <jgraham> | There is a concept of filters for treewalkers so you might be able to use the built-in sanitizer as the first stage and then write something that implements the filter interface for your own custom sanitization |
| 15:09 | <jgraham> | Not really documented though so you will need to look at the code |
| 15:10 | <jgraham> | m_ali: (as an aside I guess using cElementTree or lxml is better than using DOM as the intermediate tree format, especially if you con't plan to manipulate the tree at all and so don't care about the API) |
| 15:12 | <jgraham> | (and you can do tree = html5lib.parse(input, treebulder="lxml") rather than constructing a intermediate HTMLParser object by hand) |
| 15:12 | <jgraham> | (I think parseFragment should work as well, if not file a bug) |
| 15:14 | <m_ali> | Can't see object as an accepted element here: http://wiki.whatwg.org/wiki/Sanitization_rules or am i missing something? |
| 15:15 | <m_ali> | Also, I don't understand why I need the two stage approach? |
| 15:15 | <m_ali> | *accepted=acceptable |
| 15:18 | <jgraham> | m_ali: You can customise what elements the sanitizer accepts by changing an attribute |
| 15:18 | <mikekelly> | did someone say accept? |
| 15:18 | <jgraham> | It is harder to add extra checks on allowed attribute values |
| 15:18 | <jgraham> | That's what I am suggesting the two-stage approach for |
| 15:19 | <jgraham> | (I vaugely considered redesigning the sanitizer as a series of plugabble filters, but apart from anything else it would kill perf) |
| 15:21 | <m_ali> | hmm... ok, so do you mean that the first stage sanitize would allow <object> and my custom sanitize would then only allow those <objects> that I want? |
| 15:24 | <jgraham> | m_ali: Yes |
| 15:25 | <zcorpan_> | http://www.google.se/search?hl=sv&safe=off&client=opera&hs=yuU&rls=en&tbs=prv%3A1&q=html5test&aq=f&aqi=&aql=&oq=&gs_rfai= - google scores 62 on html5test.com |
| 15:25 | <m_ali> | thanks jgraham |
| 15:27 | <jgraham> | zcorpan_: Hah! |
| 15:28 | <jgraham> | zcorpan_: We should get them to encode the passing tests in a barcode in the area presented by google so we can determine what they pass |
| 15:28 | <zcorpan_> | 94 on beta.html5test.com |
| 15:29 | <zcorpan_> | 'no bonus points' so no more than one video and one audio codec |
| 15:35 | <zcorpan_> | http://github.com/NielsLeenheer/html5test/issues#issue/41 |
| 15:41 | <jgraham> | (also works with Acid 3) |
| 15:41 | <jgraham> | zcorpan_: Is it me or does your bug report attribute all lines to you? Not that it matters. |
| 15:43 | <Philip`> | Presumably Google is just rendering with some version of Chrome? |
| 15:43 | <jgraham> | Yeah, seems to be a reasonably old version though |
| 15:44 | <jgraham> | Is the fourth bucket on acid3 SVG? That doesn't seem to be working at all |
| 15:47 | <gsnedders> | SVG is a compile-time option for WebKit |
| 15:51 | <zcorpan_> | jgraham: github tampers with my text |
| 15:52 | <jgraham> | zcorpan_: OK |
| 16:31 | gsnedders | wonders what to do this evening |
| 16:31 | <gsnedders> | Doing something outside would be nice given the weather, but watching football seems, uh, annoying. |
| 16:33 | <jgraham> | gsnedders: Really it won't be that bad |
| 16:33 | <jgraham> | You can always bring a book |
| 16:33 | <jgraham> | and a mp3 player |
| 16:33 | <jgraham> | and headphones |
| 16:34 | <gsnedders> | Well, if I take my closed headphones, I can avoid all of the attempts to get me a girl. |
| 16:34 | <gsnedders> | But it doesn't really seem worthwhile going to a football match for that. |
| 16:34 | <Dashiva> | Get girls are a football match? |
| 16:35 | <Dashiva> | *at |
| 16:35 | <gsnedders> | Dashiva: Well, it's a women's football match, so presumably |
| 16:35 | gsnedders | doesn't have much experience of this |
| 16:35 | <Dashiva> | No field invasions |
| 16:35 | <jgraham> | gsnedders: We can make sure you don't sit with kilsmo |
| 16:35 | <gsnedders> | Dashiva: That's what my colleagues are probably planning for me. |
| 16:36 | <jgraham> | gsnedders: I'm really not, at least |
| 16:36 | <Dashiva> | If you do, make sure you streak |
| 16:36 | <jgraham> | I doubt Stina or Marlin are either |
| 16:36 | <jgraham> | honestly |
| 16:36 | jgraham | has no idea who else, if anyone, is going |
| 16:36 | <gsnedders> | jgraham: Now I'm just realizing how little clue I have about going |
| 16:37 | <gsnedders> | jgraham: Lars I thought was |
| 16:37 | <gsnedders> | *who is going |
| 16:37 | <jgraham> | (actually I'm not sure about Stina or Marlin but I know they were at least encouraged to go) |
| 16:37 | <gsnedders> | Dashiva: Just make sure to inact the second rule of streaking? |
| 16:37 | <gsnedders> | jgraham: Half the office was, as far as I can tell :P |
| 16:38 | <Dashiva> | gsnedders: The first rule of football is, kick the ball |
| 16:38 | <gsnedders> | Gah. Sit indoors and try and catch up on the last month in my diary, or go and watch football |
| 16:38 | <gsnedders> | Neither seems particuarly great. |
| 16:38 | <gsnedders> | But the latter at least takes adavantage of the fact that Sweden has been weather than Scotland. |
| 16:39 | <gsnedders> | s/been/better/ |
| 16:40 | <Dashiva> | Is it possible to not have better weather than Scotland? |
| 16:40 | <gsnedders> | Dashiva: The Polar regions. |
| 16:41 | <Dashiva> | The weather is less important when it's night half the year :) |
| 16:44 | <gsnedders> | jgraham: When are you going from the office, anyway? |
| 16:44 | <jgraham> | gsnedders: 18:30 |
| 16:45 | gsnedders | might come, and will wear less black than earlier if he's to sit around in the sun |
| 19:31 | <hsivonen> | gsnedders: Did you see Avenue Q in London? |
| 20:06 | <gsnedders> | hsivonen: No, I never actually managed to see it as I, well, ended up with too much else going on with friends there |
| 22:10 | <AryehGregor> | Why are people complaining about the chairs' decisions, when the decision policy clearly states that the only recourse is filing a formal objection? |
| 22:11 | <TabAtkins> | Because they don't understand the process, or don't care about it and just want things to go their way. |
| 22:11 | <TabAtkins> | (I, on the other hand, am bitching about the process itself.) |
| 22:12 | <AryehGregor> | Who decided that it would be a good idea to allow random people off the Internet to file change proposals that the working group has to deal with? Does any other WG work that way? |
| 22:12 | <othermaciej> | oh no, is there a giant flamewar? |
| 22:13 | <hober> | I bet Larry's new bug will go over really well with Sam... |
| 22:13 | <hober> | othermaciej: |
| 22:13 | <hober> | err, http://www.w3.org/Bugs/Public/show_bug.cgi?id=9836 |
| 22:13 | <TabAtkins> | heh, awesome. |
| 22:14 | <AryehGregor> | I'm pretty sure I remember the chairs explicitly saying something to the effect of "only strongly-held objections will be considered". |
| 22:17 | <Rik`> | This guy is paid by Adobe to contribute to HTML5 ? really ? |
| 22:18 | <othermaciej> | the exact text (still visible on the poll results page) is: "If you have strong objections to adopting this Change Proposal, please state your objections below." |
| 22:19 | <othermaciej> | I must admit that a response starting with "I'm not *strongly* opposed" sounds like the author is declaring it not to be a strong objection to me |
| 22:19 | <AryehGregor> | That does sound like a relatively reasonable inference. |
| 22:20 | <othermaciej> | and while certainly the chairs or WG could second guess them, it seems very unlikely something could be a strong objection if the person raising it says it isn't, and second guessing in such circumstances would amount to a fishing expedition |
| 22:22 | <mbrubeck> | Note that the reply read (my emphasis) "I'm not strongly opposed to the *concepts*.... *but*...." |
| 22:25 | <Hixie> | could someone explain to me why microdata and a suggestion to make UAs more accessible get split out, but <figure> and <aside> do not? |
| 22:25 | <Hixie> | i've been reading and rereading the respective "working group decisions" but can't for the life of me work out a consistent policy |
| 22:26 | <Hixie> | (that is assuming i'm even understanding the decisions correctly, which is hard given how utterly vague and wishy-washy they are) |
| 22:26 | <annevk> | they seem to be primarily based on the comments the change proposals receive |
| 22:27 | <annevk> | the one with the weakest comments against, wins |
| 22:27 | <Hixie> | so there's no consistent policy for me to apply? |
| 22:27 | <AryehGregor> | Sure there is, object very strongly to everything. |
| 22:27 | <AryehGregor> | I mean, any changes. |
| 22:27 | <Hixie> | no i mean as editor |
| 22:28 | <Hixie> | there's no policy that i can apply as editor before things get objected to |
| 22:28 | <annevk> | it depends on how well you can predict what the WG members are up to |
| 22:29 | <Hixie> | i can't even predict what the chairs are up to |
| 22:29 | <AryehGregor> | Well . . . make sure that your opinion happens to agree with the chairs'? I mean, basically what we've got is that the chairs can overrule you if people convince them that you're wrong. |
| 22:29 | <annevk> | clearly you need more data :p |
| 22:29 | <AryehGregor> | So unless you can read their minds, or pay them off or something, I don't that you'd be able to avoid ever being overruled. |
| 22:29 | <Hixie> | AryehGregor: the chairs' opinions don't agree with the chairs' |
| 22:29 | <AryehGregor> | That's a problem with most people. |
| 22:30 | <TabAtkins> | AryehGregor: Not "people convince them that you're wrong". It's "people convince them that they'll object more strenuously than people on the other side". |
| 22:30 | <TabAtkins> | It is *literally* a race to exhaustion. |
| 22:30 | <annevk> | he is not worried about being overruled, he is interested in applying rationale for approved change proposals to the rest of the specification |
| 22:30 | <AryehGregor> | Why? |
| 22:31 | <AryehGregor> | It seems like it would be easier to do what he thinks is best and then fix it if he's asked to. |
| 22:31 | <AryehGregor> | Given that there are no clear criteria being applied. |
| 22:31 | <Hixie> | i am not going to have my name on a spec that is internally inconsistent |
| 22:32 | <AryehGregor> | TabAtkins, no, they're supposed to be judging the technical merits of objections. I don't win if I say "I HATE THIS FEATURE AND WILL COMPLAIN TO THE DIRECTOR AND PICKET OUTSIDE W3C OFFICES". I have to convince them that my reasons are sound. |
| 22:32 | <Hixie> | if we're going to do something different than what i think is logical, that's fine, i just need to know what the logic is so that i can apply it consistently and make the spec internally consistent |
| 22:33 | <AryehGregor> | Hixie, are you suggesting that any form of HTML is internally consistent? Surely it's less inconsistent to change a few minor things because the chairs disagree with you, than to use entire horribly-designed APIs because IE happens to support them already. |
| 22:33 | <Hixie> | i don't mean the technology being defined, i mean the spec defining it |
| 22:33 | <TabAtkins> | AryehGregor: No, that's what you *want* to happen. That is definitely not what Sam specifically said they actually judge on. |
| 22:34 | <AryehGregor> | TabAtkins, I explicitly asked during microdata and was told something like what I just said. |
| 22:34 | <annevk> | while you guys figure this out I'm gonna read my next book |
| 22:34 | <Hixie> | i've been told they use the value of arguments too |
| 22:34 | <Hixie> | though clearly they don't |
| 22:35 | <TabAtkins> | AryehGregor: And now he's saying something different. |
| 22:35 | <othermaciej> | it seems like inclusion/exclusion is (a) a judgment call based on many factors and (b) relevant only to the snapshot of the language that will be labeled "HTML5" by the W3C (therefore some of the factors being the situation at the present moment), and therefore I am not sure what it would mean to have "internal consistency" on the set of things included |
| 22:36 | <Hixie> | othermaciej: why is <aside> in but microdata out? the arguments to both seem to apply equally |
| 22:36 | <AryehGregor> | Consistency would be an issue if the chairs were dictating exact wording, but saying a feature or paragraph here or there should be removed doesn't seem like it meaningfully increases inconsistency. |
| 22:36 | <TabAtkins> | The change proposal part of the process turns out to be a farce, unfortunately. Even when I "win" I'm angry at the results. I think I just won't contribute in the future. |
| 22:37 | <Hixie> | AryehGregor: why is one paragraph of advice in but another not? |
| 22:37 | <AryehGregor> | Hixie, the major reason for leaving out Microdata was that it should compete with RDFa on an equal footing, wasn't it? |
| 22:37 | <mbrubeck> | Hixie: That's an inconsistency in process - it doesn't necessarily lead to an inconsistent spec. |
| 22:37 | <mbrubeck> | (any more than a consistent process would necessarily lead to a consistent spec) |
| 22:38 | <othermaciej> | I don't want to get too much into it, since I'm technically on vacation still, but Tab, I am curious what you find upsetting about the results and I'll ask you about it later maybe |
| 22:38 | <Hixie> | AryehGregor: shouldn't <aside> compete on an equal footing with other proposals? |
| 22:38 | <TabAtkins> | othermaciej: Sure, just ping me whenever. |
| 22:39 | <Hixie> | TabAtkins: i feel similarly, although in my guess i'm not angry at the results, i am just baffled |
| 22:39 | <AryehGregor> | Hixie, there are no other proposals that are written up in specs and actively edited, are there? You can't say it should compete on an equal footing with something that's not in any spec and won't be put in one anytime soon. |
| 22:40 | <Hixie> | AryehGregor: so things that have active specs should be split out? i could apply that reasoning, if that's the reasoning |
| 22:40 | <AryehGregor> | It seemed fairly essential in the RDFa case. But the chairs have declined to decide anything on a general basis, only case-by-case. |
| 22:40 | <Hixie> | AryehGregor: i'd have to take out the event handler attributes (compete with XML Events), <a> and <link> (competes with XLink), rel="" (competes with RDFa and Microdata)... |
| 22:40 | <Hixie> | AryehGregor: i don |
| 22:40 | <AryehGregor> | So maybe you should ask for a change in the decision policy. |
| 22:40 | <Hixie> | er |
| 22:41 | <Hixie> | AryehGregor: i don't mind them deciding on a case-by-case basis so long as they apply their reasoning consistently to everything in the spec |
| 22:41 | <Hixie> | but they don't |
| 22:41 | <TabAtkins> | AryehGregor: Lief's <object> proposal is a competing idea. ^_^ |
| 22:43 | <jgraham> | """The decision about which objections are strong or are not strong should be |
| 22:43 | <jgraham> | working group consensus based""" |
| 22:43 | <jgraham> | We should totally have a poll for each objection |
| 22:44 | <Hixie> | consensus is a dumb way to do language design |
| 22:44 | <othermaciej> | jgraham: that suggestion reminds me of this: http://volokh.com/2010/06/01/sabotage-or-how-dilbert-won-the-war/ |
| 22:46 | <jgraham> | Yes, I think it is rather clear Larry is seeking to slow down progress |
| 22:48 | <AryehGregor> | It doesn't seem that way to me. His actions seem explainable by good faith. |
| 22:48 | <othermaciej> | To be clear, I meant your comical suggestion ("have a poll for each objection"), not Larry's |
| 22:48 | <AryehGregor> | I tend to always assume good faith. It works pretty well. |
| 22:48 | <jgraham> | AryehGregor: It also means that you are wrong most of the time |
| 22:48 | <jgraham> | Hopefully that doesn't matter |
| 22:48 | <AryehGregor> | jgraham, I beg to differ. |
| 22:49 | <jgraham> | Well I can't really demonstrate "most" |
| 22:49 | <othermaciej> | I am not sure what Larry's motivations are, and ideally I can do my job without having to make such a determination |
| 22:49 | <mbrubeck> | I'm really confused about Larry's "objection" - is it an objection to the change proposal at all, or is it really a statement of support for the change proposal? |
| 22:49 | <jgraham> | So "a non-negligible fraction" |
| 22:49 | <othermaciej> | I think most people are acting in what they themselves consider to be good faith most of the time |
| 22:49 | <mbrubeck> | oh, I see |
| 22:50 | <mbrubeck> | It objects to the "Change Proposal to Keep" not the "Change Proposal to Remove" |
| 22:50 | <jgraham> | FWIW many people consider slowing down progress to be a reasonable goal |
| 22:50 | <jgraham> | for my definition of progress |
| 22:50 | <othermaciej> | sometimes people behave in ways that are consciously insincere, but that is the exception |
| 22:52 | <othermaciej> | If anyone really had a conscious goal of "delay progress on HTML5", then participating in the HTML WG on that basis would arguably be bad faith |
| 22:52 | <jgraham> | (fwiw I don't see how you could say "ecision about which objections are strong [...] should be consensus based" and not realise that this would have the effect of slowing things down) |
| 22:53 | <jgraham> | (so it is at least true that he must believe that slower progress is acceptable) |
| 22:53 | <jgraham> | (although he could conceviably believe that it would lead to faster progress at a later time for some reason I don't see) |
| 22:53 | <mbrubeck> | He might believe both that it would slow the process and that the slower process would lead to a better ultimate outcome. That wouldn't be bad faith. |
| 22:54 | <mbrubeck> | Change is not always the same as "progress" - faster change does not magically make things better |
| 22:54 | <othermaciej> | on the other hand, seeking to improve things along dimensions that others do not care much about is good faith behavior |
| 22:54 | <jgraham> | Yes, I don't claim that he is necessarily acting in bad faith |
| 22:54 | <jgraham> | I really have no idea what his motivations are |
| 22:55 | <jgraham> | (so it makes little sense to speculate about good faith/bad faith) |
| 22:55 | <othermaciej> | even if it can be hard to distinguish at times from deliberate delay |
| 22:56 | <jgraham> | Anyway, I seem to have accidentially created a tangent from the more relevent point that the Process is driving good people away from the WG |
| 23:00 | <TabAtkins> | And thus it needs to be destroyed. |
| 23:00 | <AryehGregor> | The Process, the WG, or the W3C? :) |
| 23:02 | <othermaciej> | if the process does drive good people away then we may have to revise it |
| 23:02 | <othermaciej> | I don't know what concerns have been raised of late but I'll certainly look into them when back stateside |
| 23:02 | <TabAtkins> | DESTROY THEM ALL |
| 23:04 | <crankharder> | hey all, i've been playing around with cache manifests today and am kinda unsure about somethin |
| 23:04 | <AryehGregor> | The problem with the HTMLWG is that, unlike the WHATWG, it operates under the conceit that everyone's voice needs to be heard and it's not acceptable to just dismiss people's objections out of hand. This inevitably leads to people raising all sorts of issues that no one else cares much about, because others are forced to deal with them. |
| 23:05 | <AryehGregor> | It doesn't seem like the CSSWG operates this way, so I don't see why the HTMLWG should have to. |
| 23:05 | <AryehGregor> | As far as I can tell, in the CSSWG, random people from the Internet have no right to raise objections that anyone has to pay attention to. |
| 23:05 | <othermaciej> | it's harder to become a CSS WG member |
| 23:05 | <crankharder> | is there a way to specify a wildcard URL for the CACHE directive? |
| 23:06 | <othermaciej> | however, the CSS WG does have to at least respond to all Last Call comments in some meaningful way, and those could come from random people from the internet |
| 23:06 | <AryehGregor> | Yes, that's the problem. Everyone is theoretically equal in the HTMLWG, more or less, and that inevitably requires mounds of procedure and bickering as everyone has to have their say. |
| 23:06 | <mbrubeck> | crankharder: What CACHE directive? |
| 23:06 | <AryehGregor> | They could be responded to by one group member in a few sentences. You don't need to write lengthy counterproposals with several required sections and have a survey and so on. |
| 23:07 | <crankharder> | mbrubeck: in a cache manifest file |
| 23:07 | <Hixie> | crankharder: you mean matching everything? |
| 23:07 | <TabAtkins> | othermaciej: Harder to become a member, but not to talk on the list. |
| 23:07 | <mbrubeck> | crankharder: No, no wildcard. |
| 23:07 | <crankharder> | the NETWORK directive allows a wildcard, but not the CACHE |
| 23:07 | <crankharder> | NETWORK urls must only start with what is defined, afaik the CACHE ones must match up perfectly |
| 23:07 | <Hixie> | crankharder: the URLs in the CACHE section are fetched one by one by the browser, so i don't see how the CACHE could have a wildcard |
| 23:08 | <othermaciej> | TabAtkins: true, but random people talking on the list do not have to be taken as seriously as CSS WG members, since they do not have a direct say in what WG resolutions pass |
| 23:08 | <Hixie> | crankharder: the browser can't just make up the URLs :-) |
| 23:08 | <crankharder> | hrm |
| 23:08 | <TabAtkins> | othermaciej: Indeed. And that's somewhat more useful. I was a useful and productive contributor to the CSSWG long before I become a member. |
| 23:09 | <TabAtkins> | But had I turned out crazy, shrug. I could just have been killfiled and that's that. |
| 23:09 | <AryehGregor> | The WHATWG and CSSWG both rely on a small group of qualified people making all decisions, and merely require them to address public comments (however perfunctorily) according to no particular procedure. |
| 23:09 | <AryehGregor> | (one person, in the WHATWG's case) |
| 23:10 | <jgraham> | I'm not sure the CSS WG is the best example to follow |
| 23:10 | <othermaciej> | well, one could certainly look at whether the CSS WG is on the whole a more effective organization than the HTML WG |
| 23:10 | <TabAtkins> | Hmm, for some reason my python script is now outputting things in a bad encoding. It didn't do that a second ago, though I had basically equivalent characters. |
| 23:10 | <crankharder> | so the issue i'm having boils down to my framework (rails) appending timestamps to css/js/image files -- these timestamps are used for far-future expiration caching by the web server - it's therefore impossible to add the correct css/js/image filename to the manifest since the full name is dynamically generated |
| 23:10 | <AryehGregor> | It's the only W3C WG I've followed other than the HTMLWG. |
| 23:10 | <AryehGregor> | That's the only reason I use it for examples. |
| 23:10 | <jgraham> | It's not like it is known for being hyper efficient |
| 23:10 | <othermaciej> | I think the general consensus would be that it's not |
| 23:11 | <mbrubeck> | crankharder: Can you generate the manifest dynamically with the same timestamps? |
| 23:11 | <AryehGregor> | Perhaps it's also too big. |
| 23:11 | <AryehGregor> | It's hard to argue with the WHATWG's efficiency, and that seems to stem from the fact that only one person is responsible for making all decisions. |
| 23:11 | <crankharder> | mbrubeck: probably, but I was hoping for the wildcard solution ;) |
| 23:11 | <TabAtkins> | othermaciej: Of course, one must separate the problem of editors letting specs languish, and process making specs languish. The CSSWG suffers from the former, but the latter is what we're complaining about here. |
| 23:12 | <jgraham> | It's not clear that the CSSWG problems are due to its size |
| 23:12 | <mbrubeck> | crankharder: In fact, you should be able to generate the manifest as a static file at deployment time with appropriate timestamps - that's basically what I did for an offline app project. |
| 23:12 | <jgraham> | Although they could be I guess |
| 23:12 | <othermaciej> | in the W3C, in theory the buck stops at the Director, however he does not make any of the front-line decisions |
| 23:13 | <othermaciej> | I don't think there is evidence that HTML5 is languishing; I believe the complaints cited were about bad decision process and/or incensistent basis for decisions and/or resulting inconsistency in the spec |
| 23:13 | <othermaciej> | not about the spec failing to progress |
| 23:13 | <jgraham> | othermaciej: From a real world point of view it is doing fine |
| 23:14 | <jgraham> | othermaciej: From a W3C process POV it is nearing the end of Q2 2010 and we don't seem to be much closer to LC |
| 23:14 | <TabAtkins> | othermaciej: Bad choice of words on my part. Indeed, the problem is not anything languishing. |
| 23:14 | <jgraham> | Which was supposed to happen in Nov 1009 |
| 23:14 | <jgraham> | *2009 |
| 23:14 | <mbrubeck> | It's 1001 years late! |
| 23:14 | <jgraham> | (we're not quite that late...) |
| 23:14 | <othermaciej> | jgraham: I do wish we were further along than we are, but I also believe we are in the endgame of moving towards LC |
| 23:15 | <Hixie> | html5 is doing fine so far, the concern as i see it is that a decisions process is being applied to the spec that has only avoided causing massive damage so far by luck |
| 23:15 | <Hixie> | and by my interpreting the decisions in the least damaging way each time |
| 23:16 | <othermaciej> | I expect at least Microdata will get to LC pretty soon (I plan to make it a priority to work on that when I get back from vacation) |
| 23:16 | <mbrubeck> | Hixie: Or the process is inconsistent because it involves judgment and takes inherently unpredictable human factors into account, and can't be replaced by an algorithm as you seem to wish. |
| 23:16 | <jgraham> | othermaciej: The process that is taking us there seems to be causing attrition of people willing to contribute |
| 23:16 | <othermaciej> | and once we get one draft to LC I expect the path will be smoother for subsequent ones |
| 23:16 | <Hixie> | mbrubeck: judgement is fine, so long as it is consistently applied |
| 23:16 | <Hixie> | mbrubeck: (judgement is all that's ever been used) |
| 23:17 | <othermaciej> | jgraham: that's certainly something I would be concerned about if that is in fact the case |
| 23:18 | <mbrubeck> | Hixie: I mean, above someone suggested that microdata was split out because it has a separate spec, and you said that to be "consistent" with that "rule" you would also have to split out <a> and @rel. |
| 23:19 | <mbrubeck> | but it seems obvious that instead of a rule ("do not compete with any other specs, ever") there has to be a judgment call (which other specs are likely to matter / compete / be used?). |
| 23:19 | <Hixie> | mbrubeck: so RDFa is not likely to matter / compete / be used? |
| 23:19 | <mbrubeck> | NO |
| 23:19 | <Hixie> | rel="" and itemprop="" both compete with RDFa |
| 23:19 | <Hixie> | heck, RDFa even _uses_ rel="" |
| 23:20 | <mbrubeck> | You're still trying to turn my example of *factors to be considered* into some sort of binary rule that will always produce the same outcome. |
| 23:20 | <Hixie> | (in an incompatible manner, but that's another story) |
| 23:20 | <Hixie> | yes |
| 23:20 | <Hixie> | well not necessarily binary |
| 23:20 | <Hixie> | i'm trying to apply a rule that can predict wg decision outcomes |
| 23:20 | <mbrubeck> | And I think that's the wrong problem to try to solve. |
| 23:20 | <Hixie> | so that i can just make the spec what the wg wants without us having to waste months going through this insanely heavy-weight process for each paragraph |
| 23:21 | <mbrubeck> | heh. |
| 23:21 | <Hixie> | (and yes, we're literally doing it at the paragraph level at the moment) |
| 23:23 | <TabAtkins> | Yay, encoding issue was an encoding issue! (That is, there was no problem, the file just thought it was the wrong encoding and so was displaying incorrectly.) |
| 23:53 | <MikeSmith> | Hixie: in the W3C copy of the spec, can you please change the "This specification is available in the following formats: single page HTML, multipage HTML." to "...available in the following formats: single page HTML, multipage HTML, Author Edition." |
| 23:54 | <MikeSmith> | ...with the Author Edition text being a link to "author/" subdir |