Options:
- # Session Start: Tue Dec 02 00:00:00 2008
- # Session Ident: #html-wg
- # [00:45] * Joins: Lionheart (robin@66.57.69.65)
- # [00:50] * Quits: marcos (marcos@87.196.227.43) (Quit: marcos)
- # [00:52] * Quits: tH (Rob@129.11.83.58) (Ping timeout)
- # [01:08] * Joins: marcos (marcos@87.196.227.43)
- # [01:17] * Quits: tlr (tlr@128.30.52.30) (Quit: tlr)
- # [01:51] * Joins: MikeSmith (MikeSmith@mcclure.w3.org)
- # [01:55] <MikeSmith> ChrisWilson: you there?
- # [02:04] * Quits: Dashiva (noone@84.48.51.1) (Ping timeout)
- # [02:06] * Quits: ChrisWilson (cwilso@131.107.0.101) (Ping timeout)
- # [02:08] * Joins: Dashiva (noone@84.48.51.1)
- # [02:20] * Quits: Lionheart (robin@66.57.69.65) (Connection reset by peer)
- # [02:24] * Joins: Lionheart (robin@66.57.69.65)
- # [02:26] * Quits: adele (adele@17.203.14.144) (Quit: adele)
- # [02:46] <MikeSmith> http://crypto.stanford.edu/websec/specs/origin-header/
- # [02:46] <pimpbot> Title: Origin Header for CSRF Mitigation (at crypto.stanford.edu)
- # [02:46] <MikeSmith> (for those here who ain't seen it yet)
- # [02:47] * Quits: billmason (bmason@69.30.57.49) (Quit: Leaving.)
- # [02:49] * Quits: Hixie (ianh@129.241.93.37) (Quit: reconnecting...)
- # [02:49] * Joins: Hixie (ianh@129.241.93.37)
- # [02:51] * Quits: Hixie (ianh@129.241.93.37) (Quit: oops, config error)
- # [02:51] * Joins: Hixie (ianh@129.241.93.37)
- # [02:51] * Quits: Hixie (ianh@129.241.93.37) (Quit: leaving)
- # [02:51] * Joins: Hixie (ianh@129.241.93.37)
- # [02:55] * Quits: Hixie (ianh@129.241.93.37) (Quit: leaving)
- # [02:56] * Joins: Hixie (ianh@129.241.93.37)
- # [03:02] <MikeSmith> http://lists.w3.org/Archives/Public/www-tag/2008Nov/0125.html
- # [03:02] <pimpbot> Title: RE: ZIP-based packages and URI references into them ODF proposal from Ian Hickson on 2008-11-29 (www-tag@w3.org from November 2008) (at lists.w3.org)
- # [03:03] <MikeSmith> actually, the title for that should be something more like "Making a case for 'implementation functional specifications'"
- # [03:03] <Hixie> yeah i forgot to change the subject line
- # [03:05] <MikeSmith> Hixie: anyway, I think it was well-put. But don't seem to be any responses to it yet.
- # [03:06] <Hixie> larry said he would respond
- # [03:24] * Quits: MichaelC (Michael@128.30.52.30) (Quit: ChatZilla 0.9.84 [Firefox 3.0.4/2008102920])
- # [03:28] * Quits: Sander (svl@86.87.68.167) (Quit: And back he spurred like a madman, shrieking a curse to the sky.)
- # [03:54] * Quits: marcos (marcos@87.196.227.43) (Quit: marcos)
- # [03:55] * Joins: marcos (marcos@87.196.227.43)
- # [04:13] * Quits: MikeSmith (MikeSmith@mcclure.w3.org) (Quit: sex break)
- # [04:23] * Quits: rking3 (rking3@24.5.77.167) (Quit: rking3)
- # [04:26] * Quits: maddiin (mc@87.185.243.133) (Quit: maddiin)
- # [04:46] * Quits: Lionheart (robin@66.57.69.65) (Quit: Leaving.)
- # [04:47] * Joins: Lionheart (robin@66.57.69.65)
- # [04:48] * Quits: Lionheart (robin@66.57.69.65) (Connection reset by peer)
- # [04:59] * Quits: dbaron (dbaron@63.245.220.241) (Quit: 8403864 bytes have been tenured, next gc will be global.)
- # [05:16] * Joins: MikeSmith (MikeSmith@mcclure.w3.org)
- # [05:39] * Joins: Lionheart (robin@66.57.69.65)
- # [05:44] * Quits: heycam (cam@130.194.72.84) (Quit: bye)
- # [05:49] * Joins: dbaron (dbaron@71.204.144.136)
- # [05:54] * Quits: Lionheart (robin@66.57.69.65) (Connection reset by peer)
- # [05:58] * Joins: Lionheart (robin@66.57.69.65)
- # [06:38] * Quits: Lionheart (robin@66.57.69.65) (Quit: Leaving.)
- # [06:39] * Joins: Lionheart (robin@66.57.69.65)
- # [06:42] * Quits: Lionheart (robin@66.57.69.65) (Ping timeout)
- # [06:48] * Joins: heycam (cam@203.217.95.190)
- # [06:49] <MikeSmith> @tell hsivonen Henri, the whattf.org schema still has "optgroup = option* & optgroup* & optgroup.attrs" (optgroup allowed as a child of itself)
- # [06:49] <pimpbot> MikeSmith: You got it, player.
- # [07:05] * Quits: gavin_ (gavin@99.253.193.147) (Ping timeout)
- # [07:10] * Joins: gavin_ (gavin@99.253.193.147)
- # [07:46] * Quits: MikeSmith (MikeSmith@mcclure.w3.org) (Quit: sex break)
- # [08:00] * Joins: MikeSmith (MikeSmith@mcclure.w3.org)
- # [08:01] * Quits: MikeSmith (MikeSmith@mcclure.w3.org) (Quit: sex break)
- # [08:02] * Joins: MikeSmith (MikeSmith@mcclure.w3.org)
- # [08:34] * Quits: aroben (aroben@71.58.119.193) (Connection reset by peer)
- # [08:41] * Quits: dbaron (dbaron@71.204.144.136) (Ping timeout)
- # [08:57] <Hixie> MikeSmith: any news on all these people who were volunteering to edit parts of html5?
- # [08:58] <MikeSmith> Hixie: the same news you've heard
- # [08:58] <Hixie> crickets, huh
- # [09:00] <MikeSmith> yup
- # [09:00] <Hixie> i'm kinda pissed off at all the people asking for the spec to be split, and then nobody volunteering
- # [09:01] * Joins: aaronlev (chatzilla@92.228.77.248)
- # [09:50] * Joins: Sander (svl@86.87.68.167)
- # [09:52] * Quits: MikeSmith (MikeSmith@mcclure.w3.org) (Quit: sex break)
- # [10:04] * Joins: tlr (tlr@128.30.52.30)
- # [10:05] * Joins: ROBOd (robod@89.122.216.38)
- # [10:21] * Joins: hsivonen (hsivonen@130.233.41.50)
- # [10:21] <pimpbot> hsivonen: Sent 3 hours and 32 minutes ago: <MikeSmith> Henri, the whattf.org schema still has optgroup = option* & optgroup* & optgroup.attrs (optgroup allowed as a child of itself)
- # [10:22] * Joins: MikeSmith (MikeSmith@mcclure.w3.org)
- # [10:22] <hsivonen> MikeSmith: thanks for finding the optgroup issue
- # [10:23] <MikeSmith> hsivonen: no problem. noticed that one a while back but was reminded of it by one of Hixie's checkins today
- # [10:24] <MikeSmith> Hixie: is there anything in the HTML5 date/time microsyntax spec that intentionally violates ISO 8601?
- # [10:24] * hsivonen wishes there were single-page HTML versions of SVG specs
- # [10:24] <hsivonen> ooh. there is one for 1.2 Tiny
- # [10:25] * hsivonen uses the PDF for 1.1
- # [10:31] * Joins: tH (Rob@129.11.83.58)
- # [10:36] * Joins: zcorpan (zcorpan@88.131.66.80)
- # [10:37] <MikeSmith> hsivonen: are you aware of what differences or discrepancies RFC 3339 has with current ISO 8601?
- # [10:37] <MikeSmith> I notice that http://wiki.whatwg.org/wiki/MicrosyntaxDescriptions references ISO 8601 instead if RFC 3339
- # [10:37] <pimpbot> Title: MicrosyntaxDescriptions - WHATWG Wiki (at wiki.whatwg.org)
- # [10:37] <hsivonen> MikeSmith: I am unaware
- # [10:38] <MikeSmith> OK
- # [10:38] <anne> I thought HTML5 just followed ISO8601, just like HTML4
- # [10:38] <hsivonen> that is, I am not aware of RFC 3339
- # [10:38] <MikeSmith> hsivonen: understood
- # [10:40] <MikeSmith> anne: HTML5 doesn't explicitly reference IS0 8601, so I'm wondering if it may be that by following IS0 8601, it also follows RFC 3339
- # [10:40] <MikeSmith> and if a page like http://wiki.whatwg.org/wiki/MicrosyntaxDescriptions needs to reference a standard, would seem better that it reference a free standard if possible
- # [10:41] <pimpbot> Title: MicrosyntaxDescriptions - WHATWG Wiki (at wiki.whatwg.org)
- # [10:42] <MikeSmith> anyway, I gotta go
- # [10:42] <MikeSmith> bbl
- # [10:42] * Quits: MikeSmith (MikeSmith@mcclure.w3.org) (Quit: sex break)
- # [10:48] <Hixie> ISO8601 defines about a bazillion formats
- # [10:49] <Hixie> html5 is intended to be a subset of those formats
- # [11:48] * Quits: gavin_ (gavin@99.253.193.147) (Ping timeout)
- # [11:53] * Joins: gavin_ (gavin@99.253.193.147)
- # [12:13] * Joins: maddiin (mc@87.185.191.215)
- # [12:19] * Quits: Lachy (Lachlan@85.196.122.246) (Quit: This computer has gone to sleep)
- # [12:29] * Joins: Lachy (Lachlan@213.236.208.22)
- # [12:41] * Joins: MikeSmith (MikeSmith@mcclure.w3.org)
- # [12:44] * Quits: aaronlev (chatzilla@92.228.77.248) (Quit: ChatZilla 0.9.83-rdmsoft [XULRunner 1.9.0.1/2008072406])
- # [12:54] * Joins: aaronlev (chatzilla@92.228.77.248)
- # [13:29] * Joins: myakura (myakura@122.17.190.200)
- # [13:46] * Joins: Julian (chatzilla@217.91.35.233)
- # [13:47] <Julian> RFC 3339 vs ISO 8691 (latest): as far as I recall, the latest version of 8601 allows negative years while RFC 3339 does not
- # [13:47] <MikeSmith> Julian: thanks
- # [13:48] <MikeSmith> I wonder if the HTML5 draft allows negative years
- # [13:48] <MikeSmith> I think it does not
- # [13:49] <Lachy> MikeSmith, it doesn't
- # [13:50] <Lachy> 0001 is the earliest year allowed
- # [13:50] <Julian> I think sticking to something simple and free is the right thing to do.
- # [13:50] <Julian> If we feel that something important is missing from RFC 3339, we may want to discuss revising it (with Chris Newman)
- # [13:51] <MikeSmith> OK
- # [13:51] <MikeSmith> the HTML5 draft doesn't explicitly cite/reference either
- # [13:51] <MikeSmith> so it's not an issue there
- # [13:52] <MikeSmith> but it is a concern if we want to reference those somewhere else
- # [13:52] <MikeSmith> those being RFC 3339 or whatever the current ISO date standard is
- # [13:53] <Lachy> I'm not sure HTML5 needs to reference those specs, since HTML5 defines the syntax and parsing requirements for dates on its own
- # [13:54] <MikeSmith> well, if it were useful to have a reference anywhere, so far it seems a reference to RFC 3339 would be do-able (no need to reference any ISO spec)
- # [13:54] <Julian> now I'm concerned
- # [13:54] <Julian> is HTML5 defining it's own profile?
- # [13:55] <MikeSmith> Julian: essentially, I suppose it is
- # [13:55] <Julian> without even mentioning it's a subset of ISO8601?
- # [13:55] <Lachy> http://www.whatwg.org/specs/web-apps/current-work/#dates-and-times
- # [13:55] <pimpbot> Title: HTML 5 (at www.whatwg.org)
- # [13:56] <Julian> HTML4 refers to ISO8601 (for ins/del/@datetime, for instance)
- # [13:58] <Lachy> AIUI, ISO 8601 is more complicated than necessary for this.
- # [14:00] <Lachy> hmm, it seems that neither RFC 3339 or http://www.w3.org/TR/NOTE-datetime would be appropriate references, since HTML5 allows some things that they don't
- # [14:00] <pimpbot> Title: Date and Time Formats (at www.w3.org)
- # [14:01] <Lachy> HTML5 doesn't suffer from the y10k bug like they do, and the W3C Note doesn't support Weeks
- # [14:01] <Julian> Lachy: which feature beyond the 5 digit year issue?
- # [14:01] <MikeSmith> Lachy: any specific example of conflict with RFC 3339? just the Y2K thing?
- # [14:01] <MikeSmith> that Note can't be normatively referenced anyway
- # [14:02] <MikeSmith> what does HTML5 allows that RFC 3339 doesn't?
- # [14:02] <anne> (i haven't researched the subject) does RFC 3339 define parsing rules and error states?
- # [14:02] <MikeSmith> anne: no, it doesn't
- # [14:02] <MikeSmith> not as far as I recall
- # [14:03] <Lachy> MikeSmith, AIUI, it doesn't need to be normatively referenced, since the spec defines the syntax and parsing requirements already
- # [14:04] <Lachy> Section 3 or RFC 3339 is quite vague
- # [14:05] <MikeSmith> Lachy: understood, but I'm not just asking in reference to the current spec only
- # [14:05] <MikeSmith> but in general to what can be referenced for dates and times
- # [14:07] <Lachy> if we want a spec that can be normatively referenced, then we would need to write a new one that defines syntax, parsing and error handling rules
- # [14:08] <Lachy> I'm not sure what ISO-8601 says about parsing, but I think it only defines the syntax too
- # [14:14] <Lachy> MikeSmith, RFC 3339 also allows leap seconds, HTML5 does not
- # [14:14] <MikeSmith> ah
- # [14:17] <Lachy> MikeSmith, it also allows the T and Z to be lowercase, HTML5 does not
- # [14:18] <MikeSmith> hmm, I guess my question on that would be what rationale it has for allowing them to be lowercase
- # [14:20] <Lachy> "NOTE: Per [ABNF] and ISO8601, the "T" and "Z" characters in this syntax may alternatively be lower case "t" or "z" respectively."
- # [14:21] <Lachy> wow, it appears to be attempting to specify a normative, optional conformance criteria in a non-normative note :-)
- # [14:21] <Lachy> NOTE: ISO 8601 defines date and time separated by "T".
- # [14:21] <Lachy> Applications using this syntax may choose, for the sake of
- # [14:21] <Lachy> readability, to specify a full-date and full-time separated by
- # [14:21] <Lachy> (say) a space character.
- # [14:22] <MikeSmith> weird
- # [14:22] <MikeSmith> the BNF at http://www.apps.ietf.org/rfc/rfc3339.html#page-12 seems to only allow uppercase T and Z
- # [14:22] <pimpbot> Title: RFC 3339 (at www.apps.ietf.org)
- # [14:22] <MikeSmith> and "SO 8601 states that the "T" may be omitted under some circumstances. This grammar requires the "T" to avoid ambiguity."
- # [14:23] <Julian> MikeSmith: no. "A" in ABNF means "a" or "A",
- # [14:23] <MikeSmith> ah
- # [14:23] <MikeSmith> well, that's not good
- # [14:24] * Quits: maddiin (mc@87.185.191.215) (Quit: maddiin)
- # [14:24] <Lachy> LOL, I love when specs are contradictory :-)
- # [14:26] <Lachy> apparently, ISO 8601 allows hours in the range 00-24 instead of 00-23
- # [14:26] <MikeSmith> wonderful
- # [14:27] <MikeSmith> I think I mentioned it before, but it's quite common for open-until-after-midnight businesses in Tokyo to use dates like 26:00 and 28:00
- # [14:27] <Lachy> wtf?
- # [14:28] <tlr> "open from 5:00-28:59"?
- # [14:28] <Julian> I think you need "24" when you also allow leap seconds
- # [14:28] <Lachy> Julian, no you don't
- # [14:28] <Julian> time formats *are* hard if they are supposed to cover ecerything
- # [14:28] <Lachy> you just use 60 for the seconds
- # [14:28] <Lachy> "Standard time in the Netherlands was exactly 19
- # [14:28] <Lachy> minutes and 32.13 seconds ahead of UTC by law from 1909-05-01 through
- # [14:28] <Lachy> 1937-06-30. This time zone cannot be represented exactly using the
- # [14:28] <Lachy> HH:MM format, and this timestamp uses the closest representable UTC
- # [14:28] <Lachy> offset.
- # [14:28] <Lachy> "
- # [14:29] <MikeSmith> tlr: not usually for places open 24 hours, but instead for bars and such like - 19:00-28:00
- # [14:29] <Dashiva> And TV shows
- # [14:30] <Lachy> MikeSmith, RFC 3339 also doesn't allow Weeks, same as the W3C NOTE
- # [14:30] <Philip> The people in Japan must be better at mental arithmetic than we are
- # [14:30] <MikeSmith> Dashiva: and "fashion health" establishments
- # [14:31] <Dashiva> When you're able to do 19->7, 28->16->4 isn't so much more
- # [14:31] <Lachy> my conclusion after reading that RFC is that it's totally inappropriate as a normative reference for anything
- # [14:31] <Dashiva> Perhaps you even do 28->4 directly!
- # [14:31] <Philip> Dashiva: I use 24-hour clocks, so I don't have to do 19->7 :-)
- # [14:33] <anne> didn't know that about the Netherlands, loving it :)
- # [14:33] <Dashiva> Surely you sometimes venture into parts of the real world that still use analog clocks (or displays of such)
- # [14:33] <Lachy> In Australia, it's unfortunately common for people to use 12 hour time and there are many people that struggle with 24 hour time
- # [14:34] <Julian> lachy: for the record; I disagree with that
- # [14:34] <Philip> Dashiva: If there's an analog clock, it's easier to just look at my watch instead :-)
- # [14:34] <Lachy> Julian, that people in Australia use 12 hour time?
- # [14:34] <Lachy> or about the RFC?
- # [14:35] <Julian> the latter
- # [14:36] <Lachy> Dashiva, I can't read analog clocks very well
- # [14:37] * Joins: maddiin (mc@87.185.191.215)
- # [14:37] <Lachy> Julian, why?
- # [14:37] <Lachy> it has ambiguities, leaves things undefined and even contradictory in parts. What good is that as a normative reference?
- # [14:37] * Quits: Philip (philip@80.177.163.133) (Ping timeout)
- # [14:37] <MikeSmith> I can imagine a possible solution being for a spec to reference it with qualifying language like, e.g. "/a 'date-time' as defined in RFC 3339, with the exception that leap seconds are not allowed, and the "T" and "Z" literals must be uppercase."
- # [14:38] <Lachy> it still doesn't define parsing or error handling requirements
- # [14:38] <Lachy> and it's suggestions for handling 2 digit years seem to indicate that implementations are allowed to do as they see fit
- # [14:41] * Joins: Philip (philip@80.177.163.133)
- # [14:42] * Joins: MichaelC (Michael@128.30.52.30)
- # [14:43] <Lachy> so given that neither the RFC or NOTE-datetime cover everything used by HTML5, and that HTML5 is a subset of ISO 8601, an informative reference to 8601 would be the only reasonable alternative, if such a reference is needed
- # [14:47] * Quits: Philip (philip@80.177.163.133) (Ping timeout)
- # [14:47] <Lachy> MikeSmith, perhaps we should find a volunteer to update NOTE-datetime and put it on the REC track
- # [14:48] <Lachy> (though that would have less priority that finding volunteers for some other sections in HTML5)
- # [14:48] * Joins: aroben (aroben@71.58.119.193)
- # [14:49] <MikeSmith> Lachy: or we could just contact the current RFC 3339 editor and ask
- # [14:49] <MikeSmith> ask bout getting it update
- # [14:49] <MikeSmith> updated
- # [14:50] * Joins: Philip (philip@80.177.163.133)
- # [14:50] <tlr> Mike, C. Newman is one of the Apps area directors these days, and part of the W3C/IETF liaison group.
- # [14:51] <tlr> +1 to pinging him; maybe copy public-ietf-w3c@w3.org
- # [14:51] <MikeSmith> OK
- # [14:51] <Lachy> MikeSmith, ok, that seems reasonable
- # [14:53] <Julian> Lachy: it's arguable whether it needs to define error handling
- # [14:53] <Julian> Parsing should be umambiguous by the ABNF
- # [14:53] <Julian> If it's not, there's a bug
- # [14:53] <Lachy> Julian, no, it's not. If we want interoperability, then we have to deal with the fact that implementations will receive erroneous input. Therefore, we must define error handling
- # [14:54] <Julian> but nor necseearily in RFC 3339
- # [14:54] <Julian> nor does it need to be the same everywhere
- # [14:54] <Julian> sometimes a parse error is just a parse error
- # [14:55] <Lachy> that's a tautology
- # [14:55] <Lachy> what do you mean by it?
- # [14:56] <Julian> an invalid date can be treated as an error in the message
- # [14:56] <Julian> keep in mind that RFC 3339 is used in lots of places
- # [14:56] <Julian> error handling requirements will not be the same everywhere
- # [14:59] <Lachy> sure, and there are ways to deal with that, much like the way HTML5 parsing allows for, e.g., either aborting or applying well defined error recovery
- # [15:25] <Julian> a well-defined error recovery that implies that any sequence of characters is accepted essentially closes all future extension points.
- # [15:25] <Julian> you may think it's the right thing to do in HTML5, but rest assured that many people think different
- # [15:26] <Julian> thus, in cases like date/tiem formats, it'll be up to HTML5 to define that error handling, and also to be prepared to live with the consequences of that approach
- # [15:26] <Lachy> no, it doesn't close all extension points at all
- # [15:27] <Julian> if you accept *any* sequence of characters, oit does
- # [15:27] <Lachy> in fact, it improves the chance of adding extensions since there wouldn't be such wildly disparate implementations
- # [15:28] <Lachy> CSS had relatively well defined error handling and could process any sequence of bytes, but that hasn't stopped them defining extensions
- # [15:29] <Lachy> interoperability problems are what causes all the problems with defining extensions, since you not only have to remain relatively compatible what the original spec said, but with what each implementation does
- # [15:29] <Julian> It's the difference between "must understand" to "must ignore"
- # [15:30] <Julian> if you accept every input and parse it into a date, you do "must ignore" with respect to future extensions
- # [15:30] <Julian> that may be the right thing in many cases, but *not* always
- # [15:31] <Lachy> so, whether something is specified as must understand or must ignore is irrelevent to whether or not its well defined
- # [15:31] <Lachy> I didn't say all input must be parsed into a date. That's not what the HTML5 spec currently does either
- # [15:31] <Lachy> but all input must have a well defined result
- # [15:32] <Julian> I think that's the case in RFC 3339.
- # [15:32] <Julian> If it doesn't parse according to the ABNF, it's invalid.
- # [15:32] <Lachy> the RFC contains ambiguous suggestions for letting implementations deal with 2 and 3 digit years
- # [15:33] <Julian> precisely what part?
- # [15:37] <Julian> if you are looking at http://greenbytes.de/tech/webdav/rfc3339.html#rfc.section.3
- # [15:37] <pimpbot> Title: Date and Time on the Internet: Timestamps (at greenbytes.de)
- # [15:37] <Julian> these are hints for legacy protocols that do NOT use RFC 3339
- # [15:38] <Julian> the ABNF clearly says "4DIGIT"
- # [15:40] <anne> how can the format be extended if usage of it is free to do whatever with invalid variants?
- # [15:42] <Julian> so you're mssing a statement that if something doesn't parse it shouldn't be treated as a date?
- # [15:42] <Lachy> Julian, the 4 digit requirement clearly applies to producers. But for consumers that have to deal with 2 or 3 digit years from legacy apps, the requirements are largely undefined and implementations can do as they see fit
- # [15:43] <Julian> I disagree.
- # [15:43] <anne> Julian, that'd be one possible solution I suppose
- # [15:44] <anne> but I'm just wondering what the strategy is if applications are free to do whatever they want
- # [15:44] * Parts: zcorpan (zcorpan@88.131.66.80)
- # [15:44] <anne> I'm also wondering what the advantage of such a strategy is
- # [15:46] <Lachy> Julian, "Programs wishing to robustly deal with dates generated by such broken software should detect non-numeric decades and interpret appropriately."
- # [15:46] <Lachy> what does "interpret appropriately" mean exactly for a parser?
- # [15:49] <Julian> again, I understand Section 3 as instructions for protocols that do not usethe RFC3339 format
- # [15:49] <Lachy> it would be more appropriate if the spec did define that non-conforming dates were to be ignored
- # [15:49] <Julian> Otherwise: why would it say that before actually defining the "right thing"?
- # [15:50] <Lachy> it seems to say most of that in order to justify why 4 digits years are necessary. But it unfortunately mixes it with poorly defined implementation suggestions
- # [15:51] <anne> (if the spec would say that anything not matching the grammar is not a date and should not be treated as such then that's enough error handling for it to be used by say HTML5, afaict, if it meets the other criteria as well)
- # [15:52] <Lachy> anne, it doesn't meet needs of HTML5 for everything though
- # [15:53] <Lachy> e.g. HTML5 uses the Week format, allows 5 digit years, no leap seconds. The RFC does not
- # [16:26] <Julian> "Collect a sequence of characters in the range U+0030 DIGIT ZERO (0) to U+0039 DIGIT NINE (9). If the collected sequence is not exactly four characters long, then fail. Otherwise, interpret the resulting sequence as a base-ten integer. Let that number be the year. "
- # [16:26] <Julian> so four digit year
- # [16:28] <anne> yeah, though WF2 had five digit years
- # [16:28] <Julian> For the record: I prefer an ABNF compared to that 29-step algorithm.
- # [16:29] <Lachy> Julian, which copy of the spec did you get that from?
- # [16:30] <Lachy> the latest one on whatwg.org says "at least four characters long"
- # [16:32] <anne> seems Julian is reading some out of date version indeed
- # [16:32] <anne> there is no 29-step algorithm either anymore
- # [16:39] <Lachy> Julian, the /TR/ copy of the spec is 6 months out of date. Always look at the editors draft, either on whatwg.org or dev.w3.org
- # [16:51] * Joins: billmason (bmason@69.30.57.49)
- # [17:08] * Parts: anne (annevk@213.236.208.22)
- # [17:08] * Joins: anne (annevk@213.236.208.22)
- # [17:08] <Julian> Editor's Draft 24 October 2008
- # [17:09] * Parts: anne (annevk@213.236.208.22)
- # [17:09] <Julian> That's the version I looked at, and that one has the 29 step algo :-)
- # [17:10] * Joins: anne (annevk@213.236.208.22)
- # [17:10] <Julian> the version at whatwg.org is different, but it seems to contain as many steps as before, although with more structure
- # [17:25] <MikeSmith> hsivonen: attribute content models in whattf.org schema need to be updated (minus start/loopstart/loopend/playcount, +loop)
- # [17:26] <MikeSmith> for <audio> and <video>
- # [17:30] * Quits: aaronlev (chatzilla@92.228.77.248) (Quit: ChatZilla 0.9.83-rdmsoft [XULRunner 1.9.0.1/2008072406])
- # [17:37] <hsivonen> MikeSmith: thanks. recorded as http://bugzilla.validator.nu/show_bug.cgi?id=341
- # [17:37] <pimpbot> Title: Bug 341 - Media element attributes are stale (at bugzilla.validator.nu)
- # [17:44] * Joins: rubys (rubys@75.182.87.110)
- # [17:59] <MikeSmith> hsivonen: thanks
- # [18:00] * Quits: myakura (myakura@122.17.190.200) (Quit: Leaving...)
- # [18:00] * Joins: dbaron (dbaron@71.204.144.136)
- # [18:03] * Joins: ChrisWilson (cwilso@131.107.0.103)
- # [18:07] * Quits: Lachy (Lachlan@213.236.208.22) (Quit: This computer has gone to sleep)
- # [18:12] <Philip> Julian: Is there a nice well-defined way to extract the desired structured data, using an ABNF grammar?
- # [18:14] <Philip> like a way for a specification to say that when you're given a string, you match it against this grammar, and then the number of seconds it represents is (...something specific...)
- # [18:15] <Philip> It looks like RFC 3339 doesn't do that - e.g. it says what string matches time-secfrac, but it doesn't say anywhere that you're meant to combine it with time-second and treat it as a decimal number, as far as I can tell
- # [18:15] <Julian> Just name the ABNF production? In this case "time-second"?
- # [18:15] <Julian> Ah, I see.
- # [18:15] <Julian> I guess that's left to the reader.
- # [18:18] <Philip> In this case I suppose it's reasonably obvious to the reader, but given that there's a formal way of defining what strings match the grammar it seems like there ought to be a formal way of defining what you do once you've matched the strings
- # [18:18] <Julian> The prose could state that. I'm not sure it needs to.
- # [18:28] * Joins: Lachy (Lachlan@85.196.122.246)
- # [18:34] <Philip> Julian: I suppose I'd ideally like the specification to be automatically translatable into an implementation that gives the right structured output, because that makes it easy to tell that everything is defined sufficiently precisely
- # [18:35] <Philip> but I don't know if there's any feasible way to actually do that :-(
- # [18:37] <Julian> Philip: now this is a relatively simple case where it looks feasible
- # [18:38] <Julian> I'm not sure whether that's always the case though.
- # [18:48] * Quits: maddiin (mc@87.185.191.215) (Quit: maddiin)
- # [18:59] <Philip> Is it intentional that http://www.w3.org/html/wg/html5/ hasn't been updated since 24 October?
- # [18:59] <pimpbot> Title: HTML 5 (at www.w3.org)
- # [18:59] <Philip> (http://dev.w3.org/html5/spec/ has, though)
- # [18:59] <pimpbot> Title: HTML 5 (at dev.w3.org)
- # [18:59] <Philip> (I thought that first URL used to be the latest copy too)
- # [19:00] * Philip wonders if someone like MikeSmith knows
- # [19:00] <MikeSmith> Philip: it used to just be a rewrite to dev.w3.org
- # [19:01] <MikeSmith> I think W3C systeam subsequently disabled the rewrite
- # [19:01] <Philip> MikeSmith: Ah
- # [19:02] <Philip> MikeSmith: It seems people are using that old copy and not realising it isn't up-to-date
- # [19:02] <MikeSmith> because there is some dispute about the way we are using/abusing dev.w3.org
- # [19:02] <MikeSmith> Philip: yeah, I see that now
- # [19:02] <MikeSmith> .me contemplates just quietly restoring the rewrite
- # [19:02] <Philip> like http://lists.w3.org/Archives/Public/public-html/2008Dec/0051.html
- # [19:02] <pimpbot> Title: Re: Black-box equivalence of parsing fragments directly into context node from Boris Zbarsky on 2008-12-02 (public-html@w3.org from December 2008) (at lists.w3.org)
- # [19:03] <Philip> (and I thought I remembered the same in IRC very recently, but can't find it now)
- # [19:03] <MikeSmith> yeah
- # [19:04] <Philip> MikeSmith: Just replace the www.w3.org copy with a <meta http-equiv="refresh" content="http://dev.w3.org/..."> and then everyone will be happy
- # [19:04] <Philip> Uh, that's probably content="0;http://..."
- # [19:04] <Philip> Anyway, surely nobody would complain about that
- # [19:05] <Julian> Philip: unless people sync with wget, disconnect from the internet and then complain when they don't have the spec on the airplane :-)
- # [19:06] <anne> heh
- # [19:07] <Philip> Julian: Well, apart from all the practical problems there's nothing wrong with meta refresh!
- # [19:07] <Philip> s/practical/theoretical and practical/
- # [19:08] <MikeSmith> I just changed http://www.w3.org/html/wg/html5 to a permanent redirect
- # [19:08] <pimpbot> Title: HTML 5 (at www.w3.org)
- # [19:09] <Philip> MikeSmith: Thanks!
- # [19:09] <MikeSmith> I will probably get my ass kicked for doing it
- # [19:09] * Joins: adele (adele@17.203.14.171)
- # [19:09] <Philip> MikeSmith: I won't mind that at all, as long as the redirect stays ;-)
- # [19:10] <MikeSmith> heh
- # [19:10] <MikeSmith> I can't see it making much sense to object to it redirecting... the dev.w3.org URL is public and known anyway. so it's not really adding to load
- # [19:10] <Philip> You could redirect people to the WHATWG version
- # [19:10] <MikeSmith> the underlying issue is that dev.w3.org is a single, old, slow machine
- # [19:10] <MikeSmith> heh
- # [19:11] <MikeSmith> yeah, actually
- # [19:11] <MikeSmith> but I'm sure I'd get flak for doing that too
- # [19:11] <MikeSmith> just a matter of who I'd rather piss off
- # [19:12] <Lachy> if people just read the better quality whatwg version anyway, there wouldn't be a problem
- # [19:13] <MikeSmith> I've been told that we never should have started dev.w3.org for publishing documents, that it was intended just for hosting code
- # [19:13] <MikeSmith> but, well, its use has evolved to better fit the needs of the people using it
- # [19:13] <MikeSmith> as such systems tend to do...
- # [19:14] <Lachy> the whatwg version has more useful features than the w3 version doesn't, like the update notifier, stability labels and that cross-referencing script that lists all the links linking to a dfn
- # [19:14] <MikeSmith> the real solution is to put a better machine behind it, and/or round-robin load-balance the way www.w3.org is
- # [19:14] <Philip> And multipagination
- # [19:14] <MikeSmith> do you guys reckon your telling me something I don't already know?
- # [19:14] <Lachy> Philip, I said "useful features". the pagination doesn't count :-)
- # [19:15] <Philip> Lachy: "Don't crash my browser" is a useful feature that results from multipagination :-)
- # [19:15] <MikeSmith> anyway, I got to get some sleep
- # [19:15] <Lachy> MikeSmith, no, that was just for the log readers who would want me to backup my claim that the whatwg version is better
- # [19:17] <Lachy> Philip, which browser can't handle the single page?
- # [19:21] <Philip> Lachy: IE8b1 hated it (not sure about b2), and Opera (9.62) freezes for tens of seconds when loading it
- # [19:24] <Philip> By "tens of seconds", it turns out that I actually mean about a hundred seconds
- # [19:24] <Philip> (It's not entirely frozen - it responds to my attempts at scrolling once every few seconds, before freezing again)
- # [19:25] <Philip> and it uses 100% of a 2.0GHz core, so it's doing an awful lot of thinking
- # [19:25] <tlr> must be the html5 parser
- # [19:26] <Julian> tlr: there's some script executed onload, apparently computing intensive
- # [19:26] <Philip> It downloads and parses and renders the page pretty quickly, so that bit's not a problem
- # [19:27] <Lachy> Philip, latest internal opera desktop build doesn't seem to freeze
- # [19:28] <Lachy> it's seems about as responsive as Firefox, which also has scrolling delays while it loading
- # [19:28] * Philip tries downloading the latest build
- # [19:30] <Lachy> I can't find any public build more recent than 9.62
- # [19:31] <Dashiva> Probably on Friday, considering last week's announcement
- # [19:32] <Philip> Lachy: In Linux build 4090, it still takes about 40 seconds before it starts scrolling smoothly
- # [19:33] <Julian> MikeSmith: Can I interest you in a problem with tr.rdf?
- # [19:33] <Philip> (4090 is 6 hours old, so I guess it's reasonably up-to-date on whatever branch it's on)
- # [19:44] * Joins: Lionheart (robin@66.57.69.65)
- # [19:50] * Quits: dbaron (dbaron@71.204.144.136) (Ping timeout)
- # [20:02] * Quits: gavin_ (gavin@99.253.193.147) (Ping timeout)
- # [20:06] * Quits: ChrisWilson (cwilso@131.107.0.103) (Ping timeout)
- # [20:08] * Joins: gavin_ (gavin@99.253.193.147)
- # [20:09] * Quits: rubys (rubys@75.182.87.110) (Connection reset by peer)
- # [20:22] * Joins: dbaron (dbaron@63.245.220.229)
- # [20:32] * Joins: dbaron_ (dbaron@63.245.220.241)
- # [20:32] * Quits: dbaron (dbaron@63.245.220.229) (Quit: 8403864 bytes have been tenured, next gc will be global.)
- # [20:35] * dbaron_ is now known as dbaron
- # [20:51] * Quits: tlr (tlr@128.30.52.30) (Quit: tlr)
- # [21:01] * Joins: aaronlev (chatzilla@92.228.77.248)
- # [21:01] * Quits: Lachy (Lachlan@85.196.122.246) (Ping timeout)
- # [21:06] * Quits: marcos (marcos@87.196.227.43) (Ping timeout)
- # [21:06] * Joins: marcos (marcos@87.196.201.242)
- # [21:16] * Quits: aaronlev (chatzilla@92.228.77.248) (Ping timeout)
- # [21:17] * Joins: aaronlev (chatzilla@92.226.205.238)
- # [21:56] * Joins: maddiin (mc@87.185.205.208)
- # [22:02] <pimpbot> planet: WebKit's week - #6 <http://hanblog.info/blog/post/2008/11/28/WebKit-s-week-6>
- # [22:03] * Quits: aaronlev (chatzilla@92.226.205.238) (Quit: ChatZilla 0.9.83-rdmsoft [XULRunner 1.9.0.1/2008072406])
- # [22:11] * Quits: ROBOd (robod@89.122.216.38) (Quit: http://www.robodesign.ro )
- # [22:12] * Joins: Lachy (Lachlan@85.196.122.246)
- # [22:33] * Quits: Julian (chatzilla@217.91.35.233) (Connection reset by peer)
- # [22:55] * Quits: wilhelm (wilhelm@129.241.93.37) (Ping timeout)
- # [23:00] * Quits: gavin_ (gavin@99.253.193.147) (Ping timeout)
- # [23:01] * Joins: wilhelm (wilhelm@129.241.93.37)
- # [23:05] * Joins: gavin_ (gavin@99.253.193.147)
- # [23:07] * Joins: aaronlev (chatzilla@92.226.205.238)
- # [23:18] * Quits: aaronlev (chatzilla@92.226.205.238) (Quit: ChatZilla 0.9.83-rdmsoft [XULRunner 1.9.0.1/2008072406])
- # [23:28] * Quits: heycam (cam@203.217.95.190) (Quit: bye)
- # Session Close: Wed Dec 03 00:00:00 2008
The end :)