| 01:44 | <Hixie> | vcard.org, vcard.net, vevent.org, and vevent.net are all squatted |
| 01:44 | <Hixie> | sigh |
| 01:48 | <Dashiva> | No country code .rd either |
| 01:49 | <Dashiva> | ard.vc maybe? :) |
| 01:50 | <Hixie> | i don't like splitting words with dots like that |
| 01:50 | <erlehmann> | Domain Information |
| 01:50 | <erlehmann> | Query: ml.ht |
| 01:50 | <erlehmann> | Status: Not Registered |
| 01:50 | <erlehmann> | ;D |
| 01:51 | <Hixie> | ai guess i'll go with http://n.whatwg.org/vcard and http://n.whatwg.org/vevent |
| 01:51 | <erlehmann> | D: |
| 01:51 | <tantek> | Hixie - inventing yet another vocabulary space? |
| 01:51 | <Hixie> | no, just finding URLs for the vcard and vevent vocabularies |
| 01:52 | <Hixie> | so they can be identified in microdata and so they can be converted to rdf |
| 01:52 | <erlehmann> | ht.ml.vcard would have been sooo nice :/ |
| 01:52 | <tantek> | Hixie - http://microformats.org/profile/hcard and http://microformats.org/profile/hcalendar already define the vcard and vevent vocabularies in the context of HTML (microformats etc.) |
| 01:52 | <tantek> | and work as URLs for when converting to RDF |
| 01:52 | <tantek> | with the proper #term suffix and everything |
| 01:53 | <tantek> | no need to reinvent your own new URLs |
| 01:53 | <tantek> | e.g. http://microformats.org/profile/hcard#fn |
| 01:53 | <Hixie> | that's for hcard and hcalendar though |
| 01:53 | <tantek> | and http://microformats.org/profile/hcalendar#dtstart |
| 01:53 | <Hixie> | also, the #foo stuff works differently in microdata |
| 01:53 | <tantek> | hcard = vcard in HTML |
| 01:53 | <tantek> | hcalendar = icalendar and vevent in HTML |
| 01:53 | <Hixie> | no, hcard = vcard in class="" in HTML |
| 01:53 | <tantek> | the "h" = HTML |
| 01:53 | <Hixie> | i'm talking about microdata |
| 01:53 | <Hixie> | i mean, i don't mind reusing that URL if you want |
| 01:54 | <Hixie> | but it doesn't seem that you'd want me to |
| 01:54 | <tantek> | Hixie, I'll clarify the profiles to make it clear they're not just for class |
| 01:54 | tantek | notes a to-do |
| 01:55 | <tantek> | as the hCard 1.0.1 and hCalendar 1.0.1 separate the notion of the vocabularies from the host language extension points (e.g. class in HTML3, HTML4, and HTML5, and microdata in HTML5) |
| 01:55 | <Hixie> | k, i'll use http://microformats.org/profile/hcard |
| 01:55 | <Hixie> | seems confusing to me to have two nearly-identical vocabularies with different conversions to RDF using the same vocabulary name, but whatever |
| 01:56 | <tantek> | how will the conversions to RDF be different? |
| 01:56 | <tantek> | I'm not sure I understand |
| 01:56 | <tantek> | Hixie, if you have an "ideal" conversion to RDF already in mind, I am willing to look into re-using it. |
| 01:57 | <tantek> | I don't see any good reason for there to be two different conversions for hCard/vCard to RDF. |
| 01:57 | <tantek> | and the #foo suffix on the URLs was done deliberately to be compatible with RDF style URLs - I'd advise keeping that compatibility (unless you have a very good reason to be different). |
| 01:58 | <Hixie> | short names converted to rdf from microdata, e.g. "fn" in vcard, end up looking like "http://www.w3.org/1999/xhtml/vocab#http%3A%2F%2Fmicroformats.org%2Fprofile%2Fhcard%23fn"; |
| 01:58 | <Hixie> | s/vocab/custom/ |
| 01:58 | <tantek> | whoa that's nasty |
| 01:59 | <Hixie> | i can't see another way to prevent people from ending up using the full names when roubd-tripping |
| 01:59 | <Hixie> | round |
| 02:01 | <tantek> | is round-tripping a requirement? based on which use case? |
| 02:02 | <tantek> | exporting to legacy formats (like .vcf) inevitably will result in some loss of fidelity |
| 02:02 | <Hixie> | round-tripping isn't a requirement, it's just something that's likely to happen that it's important we don't let screw us up |
| 02:02 | <tantek> | I disagree |
| 02:02 | <Hixie> | ok |
| 02:02 | <tantek> | if there isn't a use case for it |
| 02:02 | <tantek> | then round-tripping unnecessarily compromises the design |
| 02:02 | <tantek> | both .vcf and .rdf/xml are legacy stream formats IMHO |
| 02:03 | <Hixie> | i'm talking about round-tripping through a triple store |
| 02:03 | <Hixie> | which sems highly likely to occur if rdf and microdata ever get used together |
| 02:05 | <tantek> | I don't know - do you have any existing examples of HTML (of any kind) being produced from a triple-store? |
| 02:05 | <Hixie> | sure, dbpedia.org for example |
| 02:05 | <tantek> | I'm not sure that a triple-store isn't lossy in the first place |
| 02:05 | <Hixie> | or anything metaweb produces |
| 02:06 | <tantek> | I bet the HTML produced from those is lossy compared to the HTML semantics they consumed to create the triple-store. |
| 02:06 | <Hixie> | microdata is (amongst other things) an rdf serialisation, so i don't see any reason why it'd be lossy |
| 02:06 | <Hixie> | it's not like microformats in that respect |
| 02:06 | <Hixie> | microformats -> rdf is definitely lossy, sure |
| 02:06 | <tantek> | sort of |
| 02:06 | <tantek> | no that's not lossy |
| 02:06 | <tantek> | because it works with URLs |
| 02:06 | <tantek> | for each term |
| 02:07 | <Hixie> | so what's lossy then? |
| 02:07 | <tantek> | it's the reverse that's lossy |
| 02:07 | <Hixie> | i'm very confused as to what you're saying |
| 02:07 | <tantek> | rdf -> microdata |
| 02:07 | <tantek> | the end result |
| 02:07 | <Hixie> | oh sure, arbitrary rdf to microdata can be lossy, e.g. you lose datatypes |
| 02:07 | <Hixie> | i'm talking about microdata -> rdf -> microdata |
| 02:07 | <tantek> | but once it enters a triple store, who knows what else gets added to it |
| 02:07 | <tantek> | that's my point |
| 02:08 | <Hixie> | my point is just that i don't want people writing <span itemprop="http://microformats.org/profile/hcard#fn">...</span>; and expecting it to be treated the same as <span itemprop="fn">...</span> |
| 02:09 | <tantek> | why? |
| 02:09 | <Hixie> | because that's an interop nightmare |
| 02:09 | <Hixie> | the whole point of microdata is that you don't have ambiguity of this nature |
| 02:09 | <Hixie> | you can do string compares at each stage of the way |
| 02:10 | <Hixie> | without needing any built-in knowledge of the vocabularies |
| 02:10 | <tantek> | why not just define itemprop="fn" to be a shorthand for itemprop="http://microformats.org/profile/hcard#fn"; |
| 02:10 | <tantek> | ? |
| 02:10 | <tantek> | basically, define the predefined vocabularies in terms of microformats profile URLs |
| 02:10 | <tantek> | make them equivalent |
| 02:10 | <tantek> | rather than needlessly slightly different |
| 02:10 | <Hixie> | because then you need to know the vocabulary to do anything |
| 02:10 | <tantek> | that applies to *any* predefined vocabulary |
| 02:10 | <tantek> | based on that argument you should toss all predefined vocabs |
| 02:10 | <Hixie> | no, the whole point of microdata is to fix the problem that microformats had whereby you need knowledge of the vocabulary to do _anything_ |
| 02:11 | <tantek> | which the predefined vocabularies replicate |
| 02:11 | <tantek> | ergo, drop the predefined vocabularies from HTML5 microdata |
| 02:11 | <Hixie> | what predefiend vocabularies? |
| 02:11 | <Hixie> | html5 doesn't have predefined vocabularies |
| 02:11 | <Hixie> | they've been gone for hours |
| 02:12 | <Hixie> | my point is that with microdata, you can parse any microdata into a tree, then compare it to another tree, and e.g. say whether they're the same or not, without knowing the vocabularies |
| 02:12 | <Hixie> | but if we start saying that multiple strings mean the same thing, we've lost that |
| 02:12 | <Hixie> | and that's a problem |
| 02:14 | <tantek> | "gone for hours"?!? |
| 02:14 | <Hixie> | yes, keep up :-) |
| 02:15 | <Hixie> | r4066 removed them |
| 02:15 | <tantek> | LOL - wow - ok, right then. |
| 02:15 | tantek | will re-read the section later tonight |
| 02:15 | <Hixie> | more than hours in fact, almost a day and a half :-) |
| 02:16 | tantek | has been busy writing up and running #microformats value-class-pattern date and time separation test cases. |
| 02:16 | <tantek> | (as part of the inclusion of required value-class-pattern support into hCard 1.0.1 and hCalendar 1.0.1) |
| 06:07 | Hixie | tries to work out what URL to use for vEvent |
| 06:08 | <Hixie> | i guess tantek said to use hCalendar's URL |
| 06:08 | <Hixie> | but that seems even more wrong than using hCard's |
| 06:24 | <othermaciej> | is there no such thing as hEvent? |
| 06:43 | <Hixie> | no |
| 07:34 | <tantek> | othermaciej - common misconception |
| 07:35 | <tantek> | hCalendar is based on the iCalendar vocabulary, and allows either "vcalendar" or "vevent" as root class names |
| 07:36 | <othermaciej> | where does the v come from? |
| 07:37 | <Hixie> | the iCalendar spec |
| 07:37 | <Hixie> | tantek: so what URL should I use to identify a vEvent? |
| 07:37 | <tantek> | http://microformats.org/profile/hcalendar#vevent |
| 07:38 | <Hixie> | <div itemscope itemtype="http://microformats.org/profile/hcalendar#vevent">; ? |
| 07:38 | <Hixie> | that's a little verbose |
| 07:38 | <tantek> | URLs tend to be verbose, yeah |
| 07:39 | <tantek> | hence people invented prefixes, for convenience sake, thus unwittingly creating a solution with problems worse than the original problem |
| 07:39 | <Hixie> | hm, it also means i have to come up with another way to generate the the property name URLs |
| 07:39 | <Hixie> | since http://microformats.org/profile/hcalendar#vevent#sequence isn't a valid URL |
| 07:40 | <Hixie> | http://n.whatwg.org/vevent would be much shorter |
| 07:40 | <tantek> | Hixie, if you're looking for shortening, I've partially implemented a URL shortener for microformats.org |
| 07:40 | <tantek> | coincidentally, profiles were the first set of URLs that I shortened |
| 07:40 | <Hixie> | i'm really looking for memorable rather than short |
| 07:41 | <Hixie> | and without a "#" |
| 07:41 | <tantek> | e.g. http://ufs.cc/x/hcalendar is the shortlink for http://microformats.org/profile/hcalendar |
| 07:41 | <Hixie> | not a semantically equivalent one, i hope |
| 07:41 | <tantek> | why? |
| 07:41 | <Hixie> | aliases are evil |
| 07:41 | <Hixie> | they cause implementation nightmares |
| 07:41 | <Hixie> | as we were discussing earlier |
| 07:42 | <Hixie> | identifiers should be unique |
| 07:42 | <tantek> | tough. HTTP 301. every implementation MUST support. |
| 07:42 | <tantek> | and tons of interoperable implementations do |
| 07:42 | <tantek> | e.g. search engines |
| 07:42 | <Hixie> | wait, you want people to do HTTP traffic to microformats.org every time they interpret this stuff? |
| 07:42 | <tantek> | what do you mean by "interpret"? |
| 07:43 | <Hixie> | everytime they parse a microdata blob and want to find out what it is -- if you're saying that 301 must be supported, that means they have to actually fetch this value and see what it resolves to |
| 07:43 | <Hixie> | which seems like a really expensive idea for whoever is hosting microformats.org |
| 07:44 | <Hixie> | i mean, w3.org can't handle the load for DOCTYPEs, and they have multiple mirrors and dedicated IT staff |
| 07:44 | <tantek> | nah - just like they don't have to fetch DTD URLs everytime |
| 07:44 | <Hixie> | and DOCTYPEs aren't even supposed to be fetched at all |
| 07:45 | <tantek> | welcome to the spiraling unscalable nightmare that is "distributed extensibility" |
| 07:45 | <Hixie> | it seems perfectly scalable to me if you just don't resolve the URLs |
| 07:45 | <Hixie> | and just treat them as opaque strings |
| 07:45 | <Hixie> | which is what microdata says |
| 07:45 | <tantek> | sure, you just invent another layer to make them equivalent (like OWL) |
| 07:46 | <Hixie> | why would you need anything to declare them equivalent |
| 07:46 | <Hixie> | just don't have duplicates in the first place |
| 07:46 | <tantek> | ask the OWL folks |
| 07:46 | <tantek> | in practice, duplicates happen |
| 07:46 | <Hixie> | no need to create them on purpose... |
| 07:48 | <tantek> | history and all that |
| 07:48 | <tantek> | e.g. I bet you could find numerous "aliases" for the "canonical" Dublin Core terms |
| 07:49 | <tantek> | and there was no need to create them on purpose either |
| 07:49 | <tantek> | but, it happened |
| 07:49 | <Hixie> | sure, but you're suggesting creating some on purpose |
| 07:49 | <tantek> | so, to expect that they won't happen, is against currently available data |
| 09:29 | <Hixie> | wow, no objections to the microdata vocabularies spec idea yet |
| 11:06 | <Hixie> | http://www.whatwg.org/specs/vocabs/current-work/ |
| 11:07 | <Hixie> | http://dev.w3.org/html5/mdvcard/ |
| 11:07 | <Hixie> | http://dev.w3.org/html5/mdvevent/ |
| 11:07 | <Hixie> | http://dev.w3.org/html5/mdwork/ |
| 12:38 | <Hixie> | i like how john calls the whatwg our "club" in http://lists.w3.org/Archives/Public/public-html/2009Sep/0267.html |
| 12:38 | <Hixie> | do we get a treehouse? |
| 14:17 | <Dashiva> | #whatwg-secret-tree-house-no-patents-allowed |
| 16:54 | <gsnedders> | jgraham: ping |
| 17:15 | <jgraham> | gsnedders: pong |
| 17:17 | <gsnedders> | jgraham: You remember what you agreed to a couple of days ago? |
| 17:23 | <gsnedders> | jgraham: Like, me coming over? Today? |
| 17:28 | <jgraham> | gsnedders: Yes |
| 17:31 | <gsnedders> | Gah. Caller ID doesn't work when people use other people's phones. |
| 18:05 | <TabAtkins> | Well, it took all of 5 minutes to get xampp running on my machine, including download time. I really don't understand what I was whining about before. |