This is a public chat log generated from the #semsol IRC channel.
15:58:17
bengee: the order the <link rel="alternate">s in e.g. http://bnode.org/blog/2010/01/26/microdata-semantic-markup-for-both-rdfers-and-non-rdfers causes google reader to subscribe to a feed that doesn't include the actual post
15:58:44
oha
15:58:56
you mean, re-ordering will solve the problem?
15:59:29
yes, I think putting all at the top ought to do it
15:59:56
what I always do is when I see the feed icon I copy the document URL into google reader and it seems to parse it to find the feed URL
16:00:33
so, moving meta and title down should do the trick?
16:00:52
ah, right, I see
16:00:59
the FOAF stuff comes first
16:01:21
hmm, they are generated by the cms
16:02:55
maybe I can just order them alphabetically to move "all" up
16:11:38
haha, that would work
16:12:00
also, there's an XML error in http://bnode.org/blog/atom1/all.atom (so I'll just use RSS)
16:12:56
ugh, who programmed this crap?
16:13:29
hehe, clearly someone's generating XML by doing string concatenation, the perfect recipe for disaster
16:14:07
what's the xml error?
16:17:30
ah, think I got it
16:22:53
it was the slideshare embed markup
16:32:19
ok, foolip, bnode.org should be improved now
16:33:36
the alternates are now ordered so that "all" comes first, the borken slideshare markup that broke .atom feeds is fixed
16:36:38
bengee: cool, thanks
16:36:54
thank *you*
16:36:57
you have a pending comment for that post, by the way
16:37:02
(mine)
16:37:11
foolip checks what semsol.org is
16:37:37
I've been asked for the reason why google reader doesn't work with my blog before, never had an answer
16:41:03
is it possible to subscribe to comments for an individual post, or get email notification?
16:45:51
unfortunately not
16:46:10
you know, I'm an RDFer, we are snake-oil salesmen ;)
16:48:23
a new blog engine is in the works, that'll hopefully be more Web 2.0ish
16:48:49
haha, please do continue reviewing microdata. for political reasons it hasn't gotten a lot of technical review from people who really know RDF
16:49:08
perhaps we should enhance you to snake-owl :)
16:49:17
lol
16:50:16
WRT the "works with any rdf vocab", that's indeed true, but as you point out, you can't always detect non-RDF vocabs
16:50:25
^Jenny^: not Jeni Tennison
16:50:31
oops
16:50:38
I meant to ask are you Jeni?
16:50:50
no, sorry, I'm not
16:51:12
ok, just looking for someone who knows rdfQuery :)
16:51:15
I guess I'm an underdog :)
16:51:28
just doing some shit with wikis and owl :)
16:51:56
bengee: so I guess the only way is to whitelist certain itemtypes?
16:52:03
or do you have another idea?
16:52:52
^Jenny^: do you know if it's possible to express in owl something like prefix1:$foo owl:sameAs prefix2:$foo, where $foo is any string?
16:53:08
i.e. basically saying that two prefixes are equivalent
16:53:12
re rdfQuery, she's sometimes on #swig, I think, or at least other people working with rdfQuery
16:54:01
thats not that easy... owl:sameAs works for individuals - so the string is the "identifier" of the individual
16:54:22
I'd blacklist the itemtypes that are special, like the three built-ins, and others the whatwg might define
16:54:25
that means you say prefix1:individual is the same as prefix2:individual - not prefix1 is the same as prefix2
16:55:00
e.g. semsol:jenny is the same as zfmk:jenny :)
16:55:04
I assume most vocabs will use some schema format that's in line with the algorithm, like XMDP, RDF Schema, or OWL
16:55:16
bengee: what about very old RDF vocabularies, do they really *all* work like that?
16:55:38
yes, unless you ask an RDFer who will scream "URI opacity"
16:55:56
^Jenny^: so the only way is to enumerate all members of the vocabs?
16:56:11
i.e. in *theory+ you can name your classes with any URI you like
16:56:19
bengee: exactly, URIs are supposed to be opaque, how can this really work in 100% of cases?
16:56:38
foolip: at least to my knowledge - if it works any other, tell it to me please - I need something like that, too :/
16:56:47
it's because of the first RDF syntax RDF/XML
16:56:53
with microdata there won't be any reason for people to use RDF-like vocab URIs, because they will be redundant
16:57:00
it introduced rdf:ID
16:57:11
e.g. http://n.whatwg.org/work#work would be odd
16:57:19
^Jenny^: ok :)
16:57:31
bt whole equal namespaces or "prefixes" are stupid anyway... normally they only map each other in a part-of-relationship
16:57:36
bt=but
16:58:03
rdf:ID encouraged vocab#term URIs
16:58:05
if two namespaces are absolutely equal you should ask yourself why you have two of them :)
16:58:11
^Jenny^: right, the question is mapping microdata prefixes to other vocabularies like bengee suggested (don't know if you read public-html)
16:58:47
(perhaps I should start :)
16:58:59
hehe, only if you have good mail filters
16:59:21
bengee: realistically I think people will only want to use a limited number of RDF vocabularies as-is, so is there a real problem with whitelisting?
17:00:28
RDF vocabs appear, evolve, and get abandoned much faster than other ones
17:00:55
the whole idea of RDF is working around centralized registries
17:01:18
I usually have throw-away per-app vocabs
17:02:04
that's why the whole RDF community moaned when the microformats folks suggested a wiki page for ®el values
17:02:12
@rel even
17:03:09
the question is, how many non-XMDP, non-RDFS vocabs will there be
17:03:14
some rdf vocabs are still being fiddled with 10 years later...
17:03:27
(for better or worse!)
17:03:59
and of those, how harmful will the wrongly created RDF triples be.
17:05:02
i'm yet to meet anyone hurt by a triple
17:05:06
heh
17:05:19
at some point, the precision in markup is outweighed by natural noise in our data
17:05:21
lets ask on #porno
17:05:44
- users hiding stuff, omitting stuff, making mistakes or little lies in form fields; info going out of date; version shifts, ... sites with different emphases etc
17:06:05
re your blog post, i don't think anyone seriously lobbyed activitystreams nor openid for it to be rdf based
17:06:08
and that's fine
17:06:17
there's a general attitude of getting data out there being good
17:06:37
some minority of vocal enthusiasts can be a bit boolean-minded, sure...
17:06:57
I tried to get an RDF Schema into openid.net
17:07:37
openid was already a crowded party :)
17:07:39
almost succeeded, but then the wondered how to cover v 1.1 and v2, and the thing was discontinued
17:08:00
s/the wondered/they wondered/
17:08:35
and I heard a lot of people scream activitystrea.ms were reinventing RDF
17:08:54
but, true, no real lobbying there
17:09:14
'a lot'?
17:09:36
i'm helping maintain activitystreams mailing list (spam filtering; oauth too), and have edit privs on the wiki
17:09:42
though i only edited one line so far
17:09:45
bengee: is it your itention to blacklist http://n.whatwg.org/work ?
17:09:55
i think that kind of approach is better than coming in saying 'use our stuff!'
17:10:03
just be useful and maybe at some point things will converge more
17:10:34
ok, some, not "a lot" ;)
17:10:45
bengee likes activitystreams
17:10:56
it's stronger when the activities are things in a website
17:11:03
but it's well motivated and a nice community
17:11:12
they are moving out from being atom based, to being atom-and-json-too
17:11:15
they can carry RDF updates through pubsubhubbub
17:11:19
portable contacts also straddling that
17:11:22
otherwise itemprop="work" would end up generating that very predicate -- i.e. http://n.whatwg.org/work appears both as a type and a predicate
17:11:33
oh, do you haev a pointer? i don't know the rdf / pubsubhubbub connection
17:12:12
foolip, yes, http://n.whatwg.org/* would need to be skipped by the RDF extractor
17:13:00
bengee: if you can write up the exact algorithm I can implement it in http://foolip.org/microdatajs/demo/live.html and see if it breaks stuff or not
17:13:11
oh, cool
17:13:38
I plan to put my extractor online, too. Could be interesting to compare the output
17:14:07
do you want to change <http://www.whatwg.org/specs/web-apps/current-work/multipage/converting-html-to-other-formats.html#rdf> or just to define an alternative for specialized purposes?
17:14:47
because obviously for site-specific use you can use whatever behavior you want
17:14:49
danbri, the RDF/PuSH stuff is still very secret. I'll have an app to demo it.
17:14:57
;)
17:15:20
is it native support at PuSH, or something external you made?
17:15:54
*Ideally*, I'd move the RDF mapping out of Microdata and leave that to some RDF group.
17:16:30
so that hixie doesnt have to waste time with it, and the RDF community could base it on their needs
17:17:44
if kept in the Microdata spec, I'd like to see a changed RDF algorithm, but this could also be an external post-processing step that takes the "official" RDF triples as input
17:18:19
danbri, it's transparent in PuSH
17:18:39
bengee: well, what matters is that the spec is sound and that it gets implemented
17:18:47
certainly input from RDFers would be most helpful
17:19:41
so, really just an activitystreams notation for resource updates that the receiving subscriber can translate to SPARQL Update
17:20:51
and pubsubhubbub has (low-level) private streams, so you can use it for private data exchange if the publisher is also its own hub
17:21:15
foolip, ok, great
17:21:22
i should read up on PuSH first i guess