This is a public chat log generated from the #semsol IRC channel.
11:28:08
bengee, i posted the loader script I'm using: http://danbri.org/words/2008/01/03/243
11:30:17
oh, cool
11:30:34
I have that wordpress plugin almost working, too
11:30:52
and figured out how to define sidebar widgets
11:33:05
i
11:33:21
plugins seem to be interacting and screwing things up, including somehow my feeds for planetrdf
11:33:27
but i'll be back :)
11:33:54
i wanted to ask about your smushing stuff: does it change the graph(s), or leave untouched?
11:34:33
it changes the internal identifiers
11:35:37
i.e. it doesn't directly change the graph data, but queries return different results after smushing
11:36:52
values are not lost
11:37:08
hmm, that's a bit ricky to explain..
11:37:42
there is a table that maps values (iris, literals, bnode ids) to an internal ID
11:38:09
so its like its inferring a private graph of sameAs relations?
11:38:25
the ID changing being your implementation trick for recording that...?
11:38:30
those mapping tables have (val, id, cid) where cid is "consolidated id"
11:39:24
the sql rewriter uses "cid" instead of "id", which are the same initially, but may be different after smushing
11:39:28
can you back out after smushing on dodgy data?
11:40:18
this allows you to keep using the URI or bnode ids you are used to in your queries, but still let ARC use consolidated IDs internally
11:40:55
that's cool
11:42:19
the undoConsolidation() method is still missing, I just discovered...
11:42:58
it'd just be a SET cid = id, I should add that..
11:43:23
you can currently reset the cids by re-loading a graph
11:43:58
slightly related in #jena just now, Andy just said:
11:43:59
AndyS: Blank nodes are going to be tricky. Mulgara does not provide stable blank nodes across transactions. Writing an RDF editor is "challenging". Or don
11:44:10
talking about the bridge to mulgara store they're working on
11:44:24
ah
11:44:33
what commitments does arc2 make re bnode id longevity?
11:45:00
that id/cid trick allowed to me implement bnode editing in that editing thingy
11:45:26
bnode ids are stable as long as the graph is not re-loaded
11:45:28
if the smush goes wrong, cid can just be restored to id?
11:45:36
yes
11:45:41
(I hope) ;)
11:45:54
cases where smush goes wrong: people using group weblog URIs in foaf:weblog
11:46:10
or if we'd made foaf:img inverse functional, and lots of sites had a default image url that ended up in it
11:46:13
etc
11:46:25
the inferencer ist still under construction
11:47:31
something is borken, it used to work nicely on twitter, but I tried recently, and the results loked wrong
11:47:34
looked
11:48:03
do you have twitter buddylist dumps?
11:48:33
no dumps
11:48:47
or service or mapping or whatever...
11:48:52
ok what are you talking about? :)
11:49:26
ARC supports LOAD <http://twitter.com/danbri>
11:49:43
it'll auto-detec the format and extract xfn and hcard as RDF
11:49:52
ah :)
11:51:15
danbri does
11:51:20
[danbri@ip-10-251-46-68 foaf]$ ./dataloader http://twitter.com/danbri
11:51:20
Loading data from http://twitter.com/danbri
11:51:37
bengee ducks
11:51:57
it might break your multi-graph query due to join overload...
11:53:06
how would it break it?
11:53:09
SELECT DISTINCT ?p WHERE { GRAPH <http://twitter.com/danbri> { ?s ?p ?o } }
11:53:09
btw doing
11:53:21
... I don't see anything like knows/friend/follows
11:54:11
there should be xfn:contact
11:55:11
not seeing it in http://sandbox.foaf-project.org/2008/foaf/ggg.php
11:56:15
the last two are xfn:me and xfn:contact here
11:57:52
doh
11:57:55
thanks
11:59:40
do we have any handy sources of xfn data where more exotic relationship types might be used?
11:59:54
i wanted to play with ways of defining a foaf:Group by relationship types
11:59:56
oh, the inferencer does more updates than just the cid column
12:00:01
eg. workplaceHomepage of www.w3.org
12:00:08
or xfn or family tree relations
12:02:04
something on http://microformats.org/wiki/hcard-xfn-friends-lists maybe?
12:03:40
but the services probably just use the standard props
12:09:23
i guess wordpress has built support, i could do some of my own friends?
12:09:49
have you ever looked at d2rq-ish stuff? ie wrapping sparql around vanilla sql tables
12:10:13
reason i ask is issue of crawling from wordpress (sql) into wordpress (rdf sql) ... some scope for redundancy
12:10:48
not too much, although it's interesting stuff
12:10:51
ie. a wordpress installation could be serialized out as a lot of RDF, and then loaded back into an ARC-augmented wordpress (possibly the same one, or possibly one of a neighbour, colleague...)
12:11:13
...whereas if we had a d2rq'y thing, maybe can minimise that in the case where it would be stored back in itself
12:11:19
premature optimisation tho :)
12:11:27
its cool technology
12:14:04
there is an r2d2 for rap, iirc
12:18:44
oh, interesting, i'd forgotten that
12:18:49
wonder how tightly coupled it is
13:25:05
bengee wonders how/where to make the endpoint available in the wordpress plugin
13:26:13
doesn't seem to be easy to create a dynamic page outside of wp-admin
13:45:49
ah, "template_redirect"
13:54:43
re the endpoint, ... could you make the default html form selection be "html table" not "default"?
13:55:17
html table is only available for selects
13:55:25
ah fair enough
13:55:33
also - is it possible to use GET instead of POST as an option? I guess so...?
13:55:47
the endpoint form is really just a very basic tool
13:55:58
th first version didn't even have the format selector
13:56:04
and re getting JSON results ... is there an arg for wrapping the json in a function, so it can be used in remote script tags?
13:56:20
GET works for the read-only calls
13:57:18
the format arg is "format", alternatively, you can use conneg
13:57:35
e.g. &format=json
13:57:43
but that returns pure json
13:57:59
ah, you mean somehing like jsonp
13:58:05
sorry, misunderstood
13:58:18
so random pages can load it via <script
13:58:32
...it needs my_callback( json_goes_here ) ...
13:58:37
looks like I removed that functionality, hmm...
13:59:02
I had that in ARC2, including indexed json and stuff
13:59:06
er in ARC1
14:00:03
should be easy to re-add
14:01:52
mortenf waves
14:02:07
mortenf!
14:02:20
just downloaded arc2 preview
14:02:35
saw you're into wordpress stuff atm :)
14:02:39
I hope it works with your mysql now
14:03:02
mortenf tries to remember...
14:03:28
ARC1 was 4.0.18 or higher, you had something earlier
14:03:39
right
14:03:47
i think i fixed it
14:03:56
but great, did see the req.
14:03:57
I added vesion detection now to make it work
14:04:00
:)
14:04:14
morten!
14:04:20
mortenf ought to get around to getting a new box and do ubuntu...
14:04:24
hey dan
14:04:40
and I blame danbri for the wordpress(ure)
14:04:49
sorry :)
14:04:50
heh, good job dan!
14:05:01
re new box ... i've been playing with ec2 this week
14:05:06
as many new boxes as you want!
14:05:11
yeah, saw that
14:05:24
right, not sure of the implications though
14:05:27
i've spent about 6 usd so far
14:05:51
implications -- yeah, the service is realtively immature
14:05:57
but i like the way things are going
14:06:07
and much more flexibility than most hosting companies
14:06:15
complicated: money, access, performance etc.
14:07:16
access? if things go wrong you mean?
14:07:18
mortenf is about to write an arc2 version of http://www.wasab.dk/morten/blog/archives/2005/01/04/exchange-of-named-rdf-graphs
14:07:19
yay, I got that endpoit working via WP
14:07:28
yeah, mostly, i guess
14:07:45
also, of course, there's the hw geeky factor :)
14:07:47
oh, cool idea
14:07:49
yay!
14:08:14
i've been bugging libby to get her old codepiction data back online
14:08:22
but maybe i'm just old fashioned
14:08:36
you like to own your own server?
14:08:46
i think :)
14:09:07
of course, just spent holidays replacing failed drive, so...
14:09:35
raid1d rules...
14:09:56
named graph exchange sounds great
14:10:26
would make it easier for me to test (aka stress) arc2 too :)
14:10:37
oh, shit
14:10:41
;)
14:10:42
heh
14:11:01
I need to buy some raid kit
14:11:12
i have a mountain of corpsed or suspect disks lying around
14:11:38
wonder if there's a ruby client library for sparql protocol...
14:13:59
bengee: #semsol logs available as rdf somehow?
14:14:17
... to plug into my irc log watcher :)
14:14:50
there's a sparql endpoint at http://arc.semsol.org/community/irc/sparql
14:15:22
ah, nice
14:15:53
didn't have the time to create dumps
14:16:11
even the html logs are horribly ugly
14:17:18
this is quite allright, i think
14:17:46
bengee wonders if hooking into template_redirect is the proper way to add plugin-driven pages to WP
14:17:52
if literals like in http://danbri.org/2008/xss-rdf/foaf.rdf found their way into the mail or irc endpoints, would the website be compromised?
14:18:57
... asks the man with a public DELETE-enabled endpoint ;)
14:19:31
the IRC log should escape the html
14:20:23
mortenf makes mental note to check up on that
14:20:33
not 100% sure how the triples are stored atm
14:20:46
re template_redirect: not sure there really is a "proper" way
14:21:16
i've used a filter on the_content for creating a form
14:21:37
ah
14:22:53
i've used template_redirect for other content types though, so probably a good bet
14:23:02
ok
14:23:33
it's for the endpoint which doesn't really need other wp theme stuff
14:23:46
re DELETE; heh, it's all caches of data elsewhere
14:23:51
btw, i have some unfinished arc1 work at http://bzr-dev.mfd-consult.dk/sparql-press/
14:24:03
right, that i think it's good for
14:24:43
feel free to glean and grab etc.
14:24:47
danbri remembers his list of Stuff To Ask Bengee from yesterday
14:24:54
how do we get plain text?
14:24:54
why was the endpoint a white page originally?
14:24:54
can we geet the callback arg thing for json results?
14:24:55
how do I smush? non destructively?
14:24:55
how is permissioning of updates for http://sandbox.foaf-project.org/2008/foaf/ggg.php handled?
14:25:07
oh, cool, mortenf
14:25:16
that'll be handy for the options page
14:25:48
oh, i don't remember seeing http://bzr-dev.mfd-consult.dk/sparql-press/
14:26:06
note the -dev :)
14:26:24
did you announce it? did i bang my head and forget it?
14:26:50
nope, and hope not
14:27:19
i'm quote sure it's not ready for announcement, may break stuff etc...
14:27:40
have you seen the diso project?
14:27:41
also, don't remember where i left it, might not even parse atm
14:27:46
yep
14:28:08
i mentioned sparqlpress (my old wiki notes and recent hacking here) on their list yesterday
14:29:32
mortenf looks
14:31:25
re xss in irc logs: the logger creates turtle from the message which is then sent to the store. a viewer has to make sure that the data is escaped if displayed as html
14:32:17
the trice module does it, I hope: <b>this is bold</b>
14:36:32
danbri, I
14:36:44
(the plugin is called "rdf-tools")
14:37:02
e.g. ?rdf-tools=endpoint gives you the endpoint
14:37:51
other stuff could be "?rdf-tools=xfn2foaf&url=..." etc
14:38:13
that's all without wp themeing, though
14:38:48
I'll try to create a config page in wp-admin now
14:40:26
re your qlist:
14:41:20
callback: will do that now
14:41:46
plain: you mean: don't send those damn headers?
14:44:08
its a little generic; maybe it helps if the url is related to the plugin name?
14:44:17
but sure, no biggie
14:45:27
not sure if I can catch a 404, so that I could make that /rdf-tools/endpoint or so
14:45:50
the <b> in http://arc.semsol.org/community/irc/logs/2008/01/03 is escaped
14:51:05
"plain" should work for SELECT at least
14:51:23
it'll print_r the result array
14:51:36
and sends text/plain
14:52:20
you can also comment out the sendHeaders() call in your endpoint script
14:52:42
I do that for debugging
14:56:38
plain was before we got html tables :)
14:56:51
those were the days...
14:57:04
like yesterday :)
14:57:32
does arc let me turn off remote update, delete etc?
15:00:09
"endpoint_features" in the config lets you enable selected query types, but it
15:00:39
in the knowee prototype I added another thin layer which let you specify a password
15:01:15
or you could just create two endpoints to the same store, a public one for read-only operations, ad a secret or protected one for insert/load/delete
15:04:09
or just ... , "endpoint_features" => isset($_SERVER) ? array(
15:04:51
or rather isset($_SERVER) && isset($_SERVER['SERVER_NAME'])
15:05:17
SERVER_NAME is not available when the script is called from the cl
15:06:37
fixed
15:06:46
i just removed 'em for now
15:06:49
thanks!
15:10:34
bengee, please insert #-anchors in html irc logs next time you update it :)
15:12:05
willl do
15:13:34
heading out to meet lib n damian...
15:13:37
take care...
15:15:05
see you
15:39:22
bengee back from phone
15:40:34
.. and off for a bit
15:49:35
mortenf parts...
18:50:31
bengee stops ranting about drupal's pre-processing overhead after seeing what wordpress is doing internally
18:58:43
but I found the 404 trigger, so I can offer you a "/rdf-tools/sparql" path for the endpoint now, danbri
19:38:45
ugh, and no form builder in wp..