IRC log of dig on 2012-06-17

Timestamps are in UTC.

00:22:10 [rszeno]
rszeno has quit (Quit: Leaving.)
03:26:06 [melvster]
melvster (~melvin@p4FF97E4A.dip.t-dialin.net) has joined #dig
06:17:10 [rszeno]
rszeno (~rszeno@79.114.83.161) has joined #dig
09:01:42 [bergi]
presbrey: i have already a concept in mind to use wac + tac + some new property that defines a acl pingback service. so you just send some acls like you would like to have them. the more difficult part is the user interfaces.
09:05:39 [bergi]
there are many possibilities: use the acl 1:1, find the best matching existing acl, create a new custom one, extend an existing one.
09:12:49 [bblfish]
pingback seems kind of useful here
09:13:18 [bblfish]
sounds like the kind of issue to bring up on the read-write-web community group. I think the leipzig folks would love that
09:13:37 [bblfish]
cc tramp@informatik.uni-leipzig.de
09:14:02 [bblfish]
as in fact the pingback is already meant to be in the header (especially for pictures)
09:14:47 [bblfish]
On the other hand I was thinking that if one could read the ACL
09:15:02 [bblfish]
then one should be able to figure out how to become member of the group of folks who have access
09:15:42 [bblfish]
so if the group is member of webid community group
09:15:50 [bblfish]
then one should become member of that group
09:15:55 [bblfish]
or foaf of joe, ...
09:18:22 [bergi]
there could be to uses cases: the one that you describe where the acl can be public readable and the other one where you even want to hide the existence of a resource.
09:24:53 [bblfish]
if you hide the existence, then how would anyone find out about it?
09:25:11 [bblfish]
And if they found out about it, and got so far as the header, then you'd already know their identity
09:25:19 [bblfish]
especially if you just refused them access
09:26:54 [bblfish]
so perhaps this is a question of what to return on responses that are 401
09:36:45 [bergi]
it may be different for documents and triples. but for example if you have medical data in your profile and there is a triple for a hiv test. no matter what's the object value you may not even like to show people that such a test exists.
09:39:09 [rszeno]
why somebody want to put medical data in a foaf profile?
09:39:38 [bblfish]
agree, you'd have a special medical profile
09:39:40 [melvster]
bblfish did you ever work out how to do a directory listing?
09:39:56 [melvster]
PROPFIND or GET?
09:40:02 [bblfish]
GET on /
09:40:35 [bblfish]
I think http://www.w3.org/Submission/2012/02/ have some good ideas
09:40:52 [bblfish]
I used sioc, but I think we should try to see if we agree what they propose
09:41:15 [melvster]
so which content types (MIME) are supported on a directory with GET?
09:41:24 [melvster]
or i mean, did you support
09:41:41 [bblfish]
rdf
09:41:41 [melvster]
DAV specifies text/xml and application/xml iirc
09:41:50 [bblfish]
any rdf format
09:41:51 [melvster]
rdf/xml ?
09:42:06 [bblfish]
you just have the directory return
09:42:18 [bblfish]
<> a Collection
09:42:56 [bblfish]
not sure why you need more than GET. Of course you can also POST on that resource to create new resources
09:43:26 [melvster]
PUT creates new resources, POST may or may not
09:43:38 [melvster]
assuming you have access
09:43:41 [bblfish]
with PUT you must know the name in advance
09:43:51 [bblfish]
with POST you ask for resources to be created
09:43:56 [bblfish]
without knowing the name
09:44:07 [melvster]
yes i also would like to know the rationale of PROPFIND vs GET
09:44:34 [bblfish]
( probably not good understanding of content negotiation )
09:44:54 [bblfish]
anyway, I'd go with http://www.w3.org/Submission/2012/SUBM-ldbp-20120326/
09:48:00 [melvster]
of course, great!@
09:51:40 [bblfish]
I think one should certainly check out what they are doing, and where that is moving. they have a working group
09:53:19 [melvster]
yes ive already chatted to the chairs
09:53:30 [melvster]
our work and theirs should compliment each other
09:54:36 [melvster]
is a directory a Basic Profile Container (BPC)
09:55:24 [bblfish]
yes, I think
09:55:40 [bblfish]
Containers is what the Atom people called these things
09:57:42 [melvster]
ah thanks
09:59:39 [bergi]
rszeno, bblfish: why do i have to create a second resource for that use case? in the foaf ontology is written "Person - A person." and not "foaf profile". so i see no reason to not use the same resource. on the other hand for a dissociative identity disorder it could be very usefull to link all the personalities ;-)
10:00:18 [bblfish]
you don't create two people, you create different types of documents
10:00:31 [bblfish]
<> a foaf:PersonProfileDocument . is one
10:00:48 [bblfish]
<> a med:MedicalProfileDocument is the other
10:01:02 [bblfish]
you can use the same WebID in both
10:03:21 [rszeno]
is not clear for me what is the problem when resource is hidden for some reason
10:03:59 [rszeno]
if somebody know about it in a way, he also know, in the same way how to acces it
10:05:49 [rszeno]
about medical or not medical data, yes i agree some sensitive data can be in a foaf by someone but i think is weird, :)
10:06:59 [rszeno]
i mean one can decide to put in foaf
10:08:03 [bblfish]
of course you can decide to put what you want it a foaf. But it helps to split things up into files for size issues and access control
10:09:00 [bblfish]
like you could put all your site on your front page in one gigantic web page (imagine a shop with just one huge file)
10:09:07 [bblfish]
pragmatics come into play here
10:09:50 [rszeno]
spliting as cheap inference, :)
10:10:25 [rszeno]
file level acces instead of filter triples
10:10:55 [cheater_]
cheater_ (~cheater@g229233116.adsl.alicedsl.de) has joined #dig
10:11:34 [bblfish]
yes, that's the idea of REST
10:13:35 [cheater__]
cheater__ has quit (Ping timeout: 260 seconds)
10:15:44 [rszeno]
finaly what is the answer? what solution you see?
10:15:58 [rszeno]
potential solution?
10:20:41 [melvster]
3 solutions ... 1) do it in the software like facebook 2) do it with named graphs and link (could lead to a lot of files) 3) do it with triple access control
10:20:57 [melvster]
2) is the most RESTful I think
10:24:19 [bblfish]
REST is an architectural view on things. It does not specify how you store your data - in files, in DBs or whatever
10:24:37 [bblfish]
thinking of it in terms of files is just most natural
10:24:42 [bblfish]
way to start
10:27:45 [bergi]
melvster: yes, but at the end it must be also usable. spdy was invented to speed things up, so we shouldn't slow them down to much by creating many resources. so the best from my point of view is 2.5, a middle way of 2 and 3.
10:30:05 [melvster]
yes maybe this is the way ... i think as we get more experience building code it becomes easier
10:30:34 [melvster]
my public foaf profile i sometimes make an owl sameAs to my desktop computer
10:30:51 [melvster]
i run data.fm locally too
12:55:52 [timbl]
___________________________________________________________________
12:56:08 [timbl]
8:54:7 [eror] sparql: update failed for <http://tabulator.org/wiki/tabulator/tracker/state> status=406, Not Acceptable, body length=41 for query: INSERT DATA { <http://tabulator.org/wiki/tabulator/tracker/issue#issue42> <http://www.w3.org/2005/01/wf/flow#message> <http://tabulator.org/wiki/tabulator/tracker/state#Msg1339937647642">http://tabulator.org/wiki/tabulator/tracker/state#Msg1339937647642> . <http://tabulator.org/wiki/tabulator/tracker/state#Msg1339937647642> <http://rdfs.org/sioc/ns#content> "You can
12:56:09 [timbl]
tables based on Ilaria Liccardi's code you can delete, and also optionally be prompted to add extra columns. This is used for the table-based panes, but not the query views. We could combine the code for panes and views. " . <http://tabulator.org/wiki/tabulator/tracker/state#Msg1339937647642> <http://purl.org/dc/terms/created> "2012-06-17T12:54:07Z"^^<http://www.w3.org/2001/XMLSchema#dateTime> .
12:56:09 [timbl]
<http://tabulator.org/wiki/tabulator/tracker/state#Msg1339937647642> <http://xmlns.com/foaf/0.1/maker> <http://graph.facebook.com/512908782#> . }
12:56:39 [timbl]
What causes the data.fm code in tabulator.org to return 406 I wonder?
12:57:11 [melvster]
hmm 406 ... let me look
12:58:00 [timbl]
I wonder whether tabulator.org accepts the "INSERT DATA" or is it old enough to expect "INSERT" ?
12:59:09 [melvster]
timbl: the sparql update in data.fm is still an old version of "sparql plus" that bengi put in to arc2 a while ago, and presbrey ported in
12:59:44 [melvster]
i think we're kind of waiting for rdflib to update to the sparql 1.1 spec for a more robust solution
12:59:52 [melvster]
but we could probably patch things in the mean time
13:00:35 [melvster]
https://github.com/linkeddata/data.fm/blob/master/www/wildcard/POST.php#L47
13:00:45 [melvster]
is the code that gives the 406 via HTTP POST
13:01:10 [melvster]
maybe $_content_type == 'application/sparql-query'
13:01:32 [melvster]
perhaps that header needs to be set?
13:01:47 [melvster]
im not even sure that's what's in the sparql protocol doc
13:03:36 [melvster]
ive been running about 100k monthly sparql updates on my local store via .js
13:03:55 [timbl]
406 looks like a wrong comntent type not a wont syntax on the boxdy, then?
13:04:06 [timbl]
406 looks like a wrong content type not a wont syntax on the body, then?
13:04:19 [timbl]
That's od
13:04:23 [melvster]
https://gist.github.com/2944506 <-- is some js
13:04:57 [melvster]
xhr = new XMLHttpRequest();
13:04:58 [melvster]
xhr.open('POST', file, false);
13:04:58 [melvster]
xhr.setRequestHeader('Content-Type', 'application/sparql-query');
13:04:58 [melvster]
xhr.send(data);
13:06:10 [timbl]
The code the tabulator is using is in rdflib.js
13:06:11 [timbl]
https://github.com/linkeddata/rdflib.js/blob/master/sparqlUpdate.js
13:06:33 [timbl]
xhr.setRequestHeader('Content-type', 'application/sparql-query');
13:07:27 [melvster]
the query body that i tend to use is something like:
13:07:34 [melvster]
INSERT INTO <> { }
13:07:46 [melvster]
that should work in both sparql 1.1 and sparql plus
13:07:53 [timbl]
I think saying INSERT INTO <> is terrible as <> should be the default
13:08:23 [timbl]
But the problem isn't a syntax error, the 406 means content type, but the content type looks as though it should be good
13:09:26 [melvster]
true
13:14:51 [melvster]
if (protocol.indexOf('SPARQL') >=0) {
13:15:12 [melvster]
it'll need to go into that branch rather than dav, which i assume from the debug it has
13:15:47 [melvster]
let me try inserting those triples by hand
13:23:08 [melvster]
well i could insert those triples manually to /test from http://tabulator.org/wiki/tabulator/test/read-write/
13:23:24 [melvster]
let me try from a browser
13:27:46 [melvster]
postFile( 'http://tabulator.org/wiki/tabulator/test/read-write/test2', 'INSERT INTO <> { <#a> <#b> <#c> . }' )
13:27:49 [melvster]
worked for me
13:28:01 [melvster]
if i play with the syntax it gives me a 400 Bad Request
13:28:57 [melvster]
but not 406
13:30:03 [melvster]
would indicate that somehow the content type is not getting through
13:40:08 [timbl]
thank you for trying that
13:46:47 [melvster]
yw :)
16:03:48 [presbrey]
I think you are getting:
16:03:50 [presbrey]
https://github.com/linkeddata/data.fm/blob/master/www/wildcard/POST.php#L47
16:04:01 [presbrey]
406 Content-Type Not Acceptable
19:54:16 [deiu]
deiu (~andrei@2a01:e35:8b67:4160:741c:40b5:4e2b:6d12) has joined #dig
19:54:16 [deiu]
deiu has quit (Changing host)
19:54:16 [deiu]
deiu (~andrei@unaffiliated/deiu) has joined #dig
21:04:55 [deiu]
deiu has quit (Ping timeout: 264 seconds)
21:44:03 [melvster1]
melvster1 (~melvin@p4FF96A39.dip.t-dialin.net) has joined #dig
21:44:14 [melvster]
melvster has quit (Ping timeout: 246 seconds)
22:16:17 [melvster1]
melvster1 has quit (*.net *.split)
22:22:39 [melvster]
melvster (~melvin@p4FF96A39.dip.t-dialin.net) has joined #dig
22:37:33 [bblfish]
bblfish has quit (Remote host closed the connection)
22:40:16 [bblfish]
bblfish (~bblfish@AAubervilliers-651-1-276-59.w86-212.abo.wanadoo.fr) has joined #dig
22:51:55 [bblfish]
bblfish has quit (Remote host closed the connection)
23:34:16 [melvster]
melvster has quit (Ping timeout: 244 seconds)