| 01:40 | <Lachy> | Hixie, Requiem 1.8.10 is out now |
| 01:43 | <Hixie> | d'you have a .torrent for it yet? |
| 01:44 | <Lachy> | there's on on demonoid |
| 01:45 | <Hixie> | k thx |
| 01:47 | <Hixie> | seems to be down |
| 01:48 | <Lachy> | http://www.demonoid.com/files/details/1908848/10713984/ works for me |
| 01:49 | <Hixie> | huh, works from norway |
| 01:49 | <Hixie> | and from LA |
| 01:49 | <Hixie> | just not from this network |
| 01:49 | <Hixie> | weird |
| 02:46 | <Hixie> | any bibtex people here? |
| 03:11 | <hsivonen> | Hixie: anything specific? |
| 03:23 | <Hixie> | yeah |
| 03:23 | <Hixie> | "wtf is the format" |
| 03:23 | <Hixie> | specifically, how do i quote strings in it? |
| 03:27 | <Dashiva> | {String} ? |
| 03:27 | <Hixie> | what if it contains a { or a }? |
| 03:27 | <Hixie> | (if this is defined anywhere i'd be happy to just read the docs) |
| 03:27 | <Dashiva> | I use {} or "" depending on what I need, that's all I know |
| 03:28 | <Hixie> | i'm trying to write a bibtex generator |
| 03:32 | <Hixie> | but can't find any docs on what the syntax is |
| 03:32 | <Hixie> | what are the allowed characters for the key? |
| 03:32 | <hsivonen> | Hixie: http://texlipse.cvs.sourceforge.net/viewvc/texlipse/net.sourceforge.texlipse/source/bibtex6.sablecc?revision=1.6&view=markup |
| 03:33 | <hsivonen> | Hixie: non-ASCII has hairy LaTeX escaping |
| 03:34 | <hsivonen> | can't remember how to escape } |
| 03:37 | <hsivonen> | hmm. actually, you can use non-ASCII without escaping |
| 03:45 | <Dashiva> | Isn't most things tex just "the implementation is the specification"? |
| 03:46 | <Hixie> | looks like i use quotes "like this" and if i want to embed a quote i do it "like {"} this" |
| 03:46 | <Hixie> | but i can't see how to embed a lone brace |
| 03:46 | <Hixie> | i've seen at least one book whose title was something like "The lone }" |
| 03:51 | <Hixie> | maybe {\{} ? |
| 08:01 | <jgraham> | Hixie: Without testing I would expect \{ to escape a brace in BibTeX |
| 08:22 | <Hixie> | without surrounding {}s? |
| 08:22 | <Hixie> | \" doesn't work, it has to be {"}, from what i understand |
| 08:23 | <Philip`> | \" is the syntax for putting umlauts on characters, I think |
| 08:25 | Philip` | tests things |
| 08:25 | <Philip`> | Looks like \{ doesn't work, but {\{} does |
| 08:27 | <Philip`> | Uh, I might be wrong |
| 08:38 | <Philip`> | Oh, okay, so you can't actually escape braces |
| 08:38 | <Philip`> | What you can do is \newcommand{\leftbrace}{\{} in the LaTeX file that includes the bibliography |
| 08:38 | <Philip`> | and then use \leftbrace inside the BibTeX entries |
| 08:38 | <Philip`> | (according to some documentation and some testing) |
| 08:40 | <Philip`> | (You can escape braces with \{...\} if they're matched, but you need something like \leftbrace if they're unmatched) |
| 08:42 | <Hixie> | o_O |
| 08:42 | <Hixie> | what kind of wacko language is this, anyway |
| 08:43 | <Philip`> | I blame Knuth |
| 08:46 | <Philip`> | I think the idea is it's designed to work nicely in the common case, even if that's at significant expense in certain rare cases |
| 08:46 | <Philip`> | (where "it" is both TeX and BibTeX) |
| 08:49 | <Philip`> | and I've never encountered those rare cases in BibTeX in practice, since all the strings are just title={Some Nice Simple Title} and title={Some Title with {Bits that Need Case Preserving}} and author={Jo{\~a}o} and it works easily |
| 08:57 | <Hixie> | Philip`: well hopefully nobody will ever need to convert bibliographic information on a work with a single { to bibtex... |
| 09:00 | <Philip`> | I could count the number of times I've seen a work with a single { in its title on the fingers of a hand that has no fingers |
| 09:00 | <Philip`> | so that's alright in practice |
| 09:13 | <Hixie> | i've seen one :-) |
| 09:13 | <Hixie> | but yes |
| 11:21 | annevk42 | wonders what [COOKIES] will point to in HTML5 |
| 11:46 | <hendry> | is there an HTML5 appcache example somewhere? http://twitter.com/ppk/status/1715436905 |
| 11:55 | <Rik|work> | hendry: http://www.slideshare.net/Berttimmermans/iphone-offline-webapps or http://svay.com/blog/index/post/2009/02/19/Creer-un-client-Twitter-offline-pour-l-iPhone-avec-HTML5 (in french) |
| 12:31 | <hendry> | thanks Rik|work |
| 12:31 | <hendry> | there is also http://google-code-updates.blogspot.com/2009/04/gmail-for-mobile-html5-series-using.html |
| 12:32 | <hendry> | though personally i was after some minalistic sample |
| 12:43 | <jgraham> | There is some documentation on the Mozilla MDC site but it's not quite running code |
| 14:12 | jgraham | winders if hsivonen would consider placing a link to the validator.nu html parser repository somewhere on http://about.validator.nu/htmlparser/ |
| 15:01 | <philipj> | Hixie, implementing the media resource selection algorithm when checking <source> elements should wait until the DOM is stable (script has finished) but the first set value of the src attribute is the one to use really isn't much fun. |
| 15:02 | <philipj> | I'd really prefer if step 2 ("Note: By this point, the algorithm is running asynchronously.") would queue a task to run the rest, i.e. schedule it after scripts have finished |
| 15:04 | <annevk42> | philipj, you're prolly better of sending email |
| 15:38 | <Philip`> | Hmm, a surprising number of people have contacted me about my font tool, even though I intentionally didn't give a contact email address on the site |
| 15:46 | <Lachy> | Hixie, for input type=tel, the spec says "User agents may change the punctuation of values that the user enters." - Does that affect the submitted value, or just the way in which the number is rendered? |
| 15:48 | <Lachy> | in my experience, many sites don't accept any punctuation in phone numbers. |
| 15:48 | <annevk42> | "values" points to the value concept which means afaict that it affects the submission |
| 15:49 | <Lachy> | I guess, if they use type=tel, then hopefully they will fix their back end too |
| 15:49 | <annevk42> | it would be nicer if type=tel specified some canonicalized value for submission |
| 15:50 | <Lachy> | I don't think that's really possible given the wide variety of formats used for phone numbers around the world |
| 15:50 | <Lachy> | and even formats for phone numbers in the same country, depending on whether they're land line, mobile, free call or premium rate numbers |
| 15:50 | <annevk42> | international dialing is possible |
| 15:51 | <annevk42> | surely canonicalizing phone numbers is possible too then |
| 15:51 | <Philip`> | The canonical format could be created with s/[^0-9]//g |
| 15:51 | <Philip`> | but that doesn't help when silly Americans use letters instead of numbers in their phone numbers |
| 15:51 | Philip` | wonders if other nationalities do that too |
| 15:52 | <annevk42> | though some phone numbers are bound by country |
| 15:52 | <Philip`> | Also that wouldn't work when people write "+44 (0)1234 567890" |
| 15:53 | <Lachy> | IIRC, the ITU defines official phone number standards, and says something about the format. But that requires the country code to be known, and users don't always enter that into forms |
| 15:54 | <Lachy> | Philip`, is the (0) like the area code that's used when you don't explicitly dial the country code? |
| 15:56 | <hsivonen> | Hixie: I'm told you can backslash-escape curly braces in .bib |
| 15:57 | <jgraham> | Lachy: Yes |
| 15:58 | <Philip`> | hsivonen: Only works if they're matched pairs, according to the compiler's error output |
| 15:59 | <Lachy> | jgraham, so to dial my UK mobile number from a UK phone, I dial 0 7798 526 965. Or from an international phone using: +44 7798 526 965 ? |
| 15:59 | <jgraham> | Lachy: Yes |
| 16:00 | <jgraham> | Assuming that is your real mobile number :) |
| 16:00 | <Lachy> | it is |
| 16:00 | <Lachy> | it's a prepaid SIM card I use whenever I visit London |
| 16:00 | <Lachy> | it's cheaper than paying international rates with my Norwegian or Australian SIMs |
| 16:01 | jgraham | wonders when Lachy will start getting abusive calls from Mr Last Week :) |
| 16:01 | <Dashiva> | Dramatic readings of select IRC quotes |
| 16:01 | <Lachy> | my phone number has been published on my website for years. Haven't received any abusive phone calls yet |
| 16:01 | <Philip`> | Maybe nobody abusive reads your website |
| 16:02 | <Lachy> | I'm expect MLW will go looking for it now that I've mentioned it's there :-) |
| 16:02 | <Philip`> | Not that I'm implying the people reading IRC logs (hello out there!) are abusive |
| 16:02 | <Philip`> | (Not that I'm implying they're not, either) |
| 16:02 | <Dashiva> | You're implying you're not implying anything |
| 16:03 | jgraham | is confused by the implications of that |
| 16:03 | <Lachy> | Philip`, some people consider the abusive ones to be those of us in this channel |
| 16:04 | Philip` | refrains from stating an abusive comment about such people |
| 16:04 | <jgraham> | Anyway, the point is that it is amusing to imagine MLW in a phone box somewhere in rural England, feeding his 10ps into the machine to make crank calls to Lachy, whilst using a vocoder to disguise his voice |
| 16:05 | <Lachy> | he would have to know when I'm going to be in the UK for that to be effective |
| 16:05 | <gsnedders> | I'm sure you'll give away that information in here |
| 16:05 | <Lachy> | I haven't yet |
| 16:06 | <jgraham> | Lachy: Not now that he knows that your phone number is on your website. Plus you often mention that here, I think |
| 16:06 | <Lachy> | (I gave it away elsewhere already though) |
| 16:06 | <Philip`> | He could try phoning you every day |
| 16:06 | <jgraham> | He could just leave voicemail |
| 16:06 | <Lachy> | I don't think I set up my voice mail on that SIM yet |
| 16:07 | <Lachy> | I haven't set up voice mail on my norwegian number yet either, but that's cause I don't know how and can't find out cause the instructions are all in norwegian |
| 16:08 | <Philip`> | I suppose the problem with all these ideas is that they would merely be abuse against Lachy, and couldn't be justified as "GUERILLA JOURNALSIM" regardless of whether he deserves the abuse |
| 16:08 | <Lachy> | I'll be in London this weekend, arriving Saturday and leaving Sunday, on my way to Australia. |
| 16:08 | <Lachy> | I'm looking forward to receiving abusive calls now :-) |
| 16:09 | <Philip`> | (and a respectable figure like MLW wouldn't stoop to unjustified abuse) |
| 16:09 | gsnedders | has some Lachy mobile number in his phone |
| 16:09 | <Lachy> | gsnedders, you probably have my Norwegian number |
| 16:09 | <jgraham> | Philip`: No but they would be Crushing the Cabal! |
| 16:10 | <Dashiva> | You can't crush the cabal |
| 16:10 | <gsnedders> | Lachy: Yeah, it's Norwegian |
| 16:10 | <Dashiva> | At worst you get 52 pickup |
| 16:10 | gsnedders | resists temptation to call Lachy |
| 16:11 | Lachy | finds gsnedders' number in his address book |
| 16:11 | <jgraham> | gsnedders: That is probably good because the cost benefit of proving that you can use a phone vs international calling rates seems bad |
| 16:11 | gsnedders | has already found Lachy's |
| 16:12 | <gsnedders> | jgraham: Exactly. |
| 16:17 | <Lachy> | http://en.wikipedia.org/wiki/E.164 |
| 16:17 | <Lachy> | that defines the format for phone numbers |
| 16:20 | <Lachy> | The pattern attribute probably shouldn't be allowed to apply to type=tel given that browsers are allowed to automatically adjust the punctuation |
| 16:20 | jgraham | is quite concerned at the idea of browsers automatically adjusting punctuation |
| 16:21 | <Lachy> | I'm a little concerned about it. But it's what my phone does when I enter a number into the address book |
| 16:21 | <Lachy> | and, for the most part, it seems to apply country specific formatting conventions fairly reliably |
| 16:22 | <Philip`> | Is there some list of the use cases for entering phone numbers? |
| 16:22 | <Dashiva> | Many registration forms, alas |
| 16:22 | <jgraham> | I'm not sure how browsers will be able to tell what punctuation is required. Browsers will need to be compatible but there is no spec |
| 16:22 | <Lachy> | Here's a use case: A site wishes to harvest phone numbers for use by telemarketers, and needs a convenient way for users to quickly and easily enter their phone number |
| 16:22 | <Philip`> | like, is it for e.g. address books where you want a nice human-readable (and not machine-readable) format, and it's basically plain text but you want to stop people typing in garbage? |
| 16:23 | <Philip`> | or is it for machine-readable numbers so a machine can automatically phone you, without any human involvement? |
| 16:24 | <jgraham> | Philip`: Probably both |
| 16:24 | <Philip`> | In the former case I guess you'd want to allow numbers like "01234 567890 (switchboard; ask for John Smith)" too |
| 16:24 | <Lachy> | I think the use case is because it allows for integration with the user's address book, and in mobile browsers with touch screens, focussing the control gives a keypad designed for entering numbers, rather than a normal text entry keypad |
| 16:24 | <jgraham> | (and quick access to tehir contacts list) |
| 16:24 | <Philip`> | Ah, I guess the autofill case is independent of any validation or canonicalisation |
| 16:25 | <Philip`> | s/autofill/fancy input method/ |
| 16:25 | <Lachy> | I don't think we can stop people typing in garbage, because the phone number conventions use a wide variety of strange punctuation characters, inluding + ( ) # , . (space) and many others |
| 16:25 | <jgraham> | It is just a hint to a UA; the server has to be able to deal with the possibility of a garbale value anyway |
| 16:25 | <jgraham> | *garbage |
| 16:26 | <jgraham> | So having UAs mung the actual value seems bad |
| 16:27 | <Lachy> | for a normal desktop browser, yes. |
| 16:27 | <Lachy> | but for a user with a browser on a mobile phone, it could be convenient |
| 16:27 | <jgraham> | Lachy: Howso? |
| 16:28 | <Lachy> | so that the number gets formatted nicely for easier reading, without the user having to manually enter characters that aren't so easy to type on a mobile keypad |
| 16:29 | <jgraham> | Lachy: That seems to be something that is between the UA and the user if they want to have a special UI for entering numbers that pretty prints the result before updating the control |
| 16:30 | <jgraham> | Not really something that should be in the spec |
| 16:30 | <Lachy> | the spec just says it's permitted, not that it's required |
| 16:31 | <jgraham> | Lachy: The spec doesn't say every time that a UA is permitted to innovate the UI |
| 16:31 | <Philip`> | The spec normatively states that it's permitted, which seems unusual for UI issues |
| 16:31 | <Lachy> | yeah, true |
| 20:53 | <jgraham> | Hixie: Didn't zcorpan just propose that < be disallowed in _unquoted_ attributes? |
| 20:54 | <Hixie> | oh |
| 20:54 | <Hixie> | i may have misunderstood |
| 20:55 | <Hixie> | i guess that could work |
| 20:55 | <Hixie> | though if we keep adding more things that aren't allowed, it's gonna make it hard to know when to quote again |
| 20:55 | <Philip`> | That's easy - always quote |
| 20:56 | <Philip`> | You already have to pretty much do that in order to ensure compatibility with legacy clients |
| 20:56 | <Philip`> | (You need quotes for at least something like [\x00-\x1f"'`] I think) |
| 20:56 | <Philip`> | (and >) |
| 21:04 | <jgraham> | It saddens me to say that I increasingly find it hard to defend making unquoted attributes conforming |
| 21:04 | <jgraham> | Because the quotes are ugly and unnecessary |
| 21:04 | <jgraham> | Except when they are really important |
| 21:05 | <jgraham> | And the rules are rather more complex then "you just need to quote when there is whitespace or a quote character" |
| 21:08 | <Philip`> | I guess it's easy enough if you're writing it by hand, because you won't have funny characters; but if you're writing a serialiser, you either need to be extremely careful (else you'll have bugs like html5lib has, and those bugs could become security issues) or else always quote everything |
| 21:11 | <jgraham> | Philip`: Does html5lib have bugs relative to the spec or just relative to legacy clients? |
| 21:12 | <Philip`> | jgraham: Just the latter, I think |
| 21:12 | <Philip`> | (http://code.google.com/p/html5lib/issues/detail?id=92 and http://code.google.com/p/html5lib/issues/detail?id=93 ) |
| 21:13 | <Hixie> | serialisers should just quote everything unless they're trying to save bytes |
| 21:14 | <jgraham> | Philip`: Yeah I know about those issues. I guess we should quote the exra characters by default. Or maybe just always quote |
| 21:14 | <jgraham> | Hixie: Does saving bytes really make a compelling argument, especially given gzip? |
| 21:15 | <gsnedders> | I think we should quote them by default when no-quotes, but also quote everything by default |
| 21:15 | <Hixie> | jgraham: for some things, yes. |
| 21:16 | <jgraham> | Hixie: For example? |
| 21:16 | <Philip`> | If serialisers provide a make_output_more_efficient option, people are bound to select that option |
| 21:16 | <Hixie> | jgraham: if you're trying to fit something into one packet, and with gzipping you're one byte over, for instance. |
| 21:16 | <Philip`> | and even if most people don't select that option, some will, so the serialiser still has to carefully reverse-engineer legacy UAs in order to determine what it still needs to quote in order to prevent XSS attacks |
| 21:17 | <Philip`> | (because the spec doesn't provide any guidance) |
| 21:17 | <Philip`> | Hixie: In that case, you should just gzip harder! |
| 21:20 | <jgraham> | Hixie: That doesn't seem like a case that falls in the 80 part of 80/20. And in particular it doesn't seem probable that there will be a large number of pages that can't be made to fit into a single packet with quotes but do fit in without quotes |
| 21:22 | <Philip`> | And those pages can easily sacrifice validity anyway |
| 21:22 | <Hixie> | i agree that it's not a huge deal for most people, but i think making it possible is important enough. |
| 21:23 | <Hixie> | Philip`: if we're saying they are allowed do it, then we're saying it shouldn't be invalid. |
| 21:23 | <Hixie> | that's what "invalid" means |
| 21:23 | <Philip`> | Hixie: We're not saying they're allowed to do it |
| 21:23 | <Philip`> | but they can do it anyway so that's alright |
| 21:24 | <Philip`> | just like we're not saying they're allowed to use <center> simply because it's fewer bytes than <div style="align:center">, but they can do it anyway if they really care about saving bytes |
| 21:25 | <Hixie> | conformance is the technical equivalent of law (though without enforcement) |
| 21:26 | <Hixie> | just as it's not valid to say "well stealing is against the law but they can still do it if you need to", it's not valid to say "attributes must be quoted but they can still ignore that if they need to" |
| 21:26 | <Philip`> | Law without enforcement is just guidelines |
| 21:27 | <Hixie> | either we are ok with someone doing something, and we make it conforming, or we have good reasons to not allow it, and we want nobody to do it. |
| 21:28 | <Philip`> | That seems quite a valid thing to say - maybe my house is burning down so I can steal a bucket of water from a neighbour, even though it's against the law, and that's fine because there's nothing physically preventing you from doing that in extreme circumstances |
| 21:29 | <Hixie> | i disagree; i'd expect the law to allow such extremes explicitly, just like the law allows people to break other people's ribs legally in certain condititions (e.g. doing CPR to save their life) |
| 21:30 | <jgraham> | Well as Philip` says we disallow <center> even though it is shorter than <style>.a{text-align:center}</style><span class="a"> because the argument that it leads authors to do harmful things is stronger than the argument that people who need to save bytes should be allowed to write <center> |
| 21:32 | <jgraham> | (even though the argument about device-independent markup seems to be somewhat theoretical whereas the argument about quoting can be practically demonstrated) |