This is a public chat log generated from the #semsol IRC channel.
10:19:17
i got a bug report for latest version of arc2
10:30:38
i have a simple script where a remote store doesn't work with a query that has "bif:contains" - it works with other queries.
10:30:39
I tracked down the problem up to a point that I discovered the romote store doesn't have the query in $p->getQueryInfos()
10:31:55
which namespace URI are you using for the bif: prefix ?
10:33:37
bengee: none. which one should i use? it worked w/o one in the dbpedia web ui
10:33:57
ARC will need a valid PREFIX definition as the query is parsed once by the remote store in order to detect the type
10:34:31
bengee doesn't know the URI for bif:, MacTed might
10:37:58
the thing is, when I insert a simple $qt="select" into RemoteStore#s runQuery function, it works.
10:39:34
yeah, you're bypassing the test then
10:46:00
bengee tests locally
10:50:44
yeah, ARC can't parse the query because of the lacking bif: prefix
10:50:56
it's not a bug, though
10:55:52
bengee: the problem is that bif prefix isn't publicly known?
10:55:59
google for "prefix bif" doesn't bring anything helpful
10:56:03
right
10:56:22
you'll have to poke the virtuoso guys
10:56:32
they should know
10:58:01
they don't have a irc channel, do they?
10:58:07
dunno
10:58:17
otherwise. can you point me to the right forum her ehttp://boards.openlinksw.com/support/
10:58:47
i imagine you can find them in #swig
10:59:33
sorry thosh, no idea how they handle support, I have kidehen blacklisted due to his perma-spam
11:00:09
haha
11:00:58
any idea what their nicks on #swig might be?
11:02:27
you could ask on the w3c semantic-web list. I'm sure others tried to figure out the bif: namespace URI as well
11:04:01
can't you just get it back from the data from the dbpedia endpoint?
11:06:20
kwijibo: i wouldn#t know how, i just get the result sets
11:06:54
well, you must be using the bif namespace in the query?
11:07:05
change the select to a describe
11:07:13
and if the query has bif:foo
11:07:31
kwijibo, bif:contains is a "magic" predicate for fulltext search
11:07:32
look for a uri with foo at the end of it in the rdf you get back
11:07:38
ah
11:07:40
there are no triples with it
11:07:41
sorry
11:07:43
my mistake
11:07:53
and deep deep ignorance
11:07:58
;)
11:16:29
ok, i posted the question to an open link forum
12:30:36
Hi room, hey bengee.... just trying out the lasest release but I'm it still looks like I am having problems with the maths in a filter :/
12:30:54
hmm
12:31:27
thosh, bengee -- bif: prefix is Virtuoso-specific
12:33:01
arcnoob, same query as sent to the list?
12:33:14
yer
12:33:29
the sql looked ok to me
12:34:02
just wondering where to dig around in the code
12:34:42
MacTed, thosh needs the bif prefix URI so that the query can be passed through other SPARQL engines/proxies
12:36:31
arcnoob, that's not easy to nail down, but the component is the SelectQueryHandler, and the most relevant method is getFunctionExpressionSQL
12:37:18
kk thanks... I'll take a poke around :)
12:37:52
if you have an endpoint running, the latest revision allows you to see the generated SQL, which might help
12:38:38
cool, i'll take a look
12:38:51
add "endpoint_enable_sql_output" => 1 in the config array
12:39:42
and you'll see an "SQL" entry in the "output format" drop-down
12:40:11
bengee - *nods* I'm not sure that URI exists, or that such passthru is possible. doc excerpt --
12:40:13
[[[
12:40:14
A SPARQL expression can contain calls of Virtuoso/PL functions and built-in SQL functions in both the WHERE clause and in result set. Two
12:40:15
namespace prefixes, bif and sql are reserved for these purposes. When a function name starts with bif: namespace prefix, the rest of name
12:40:17
is treated as a name of SQL BIF (Built-In Function). When a function name starts with sql: namespace prefix, the rest of name is treated
12:40:18
as a name of Virtuoso/PL function owned by DBA with database qualifier DB, e.g. sql:example(...) is converted into DB.DBA."example"(...).
12:40:19
]]] - http://docs.openlinksw.com/virtuoso/rdfapiandsql.html#rdfsqlfromsparql
12:41:01
thx, MacTed
12:41:26
thosh, did you try assigning a random URI to the bif: prefix?
12:41:52
maybe the prefix def is ignored by virtuoso
12:42:43
better answer will probably come through the forum (http://boards.openlinksw.com/), or through virtuoso-users mailing list (signup at http://sf.net/projects/virtuoso)
12:42:55
(my knowledge isn't always the deepest, as much as I keep digging)
12:43:05
bengee with a random prefix i don't get an error, but also no results
12:43:43
but now I must join the traffic commute. let me know what comes of this?
12:43:44
MacTed: thx, i already posted the question here http://boards.openlinksw.com/support/viewtopic.php?f=12&t=208
12:43:46
bengee gets an error if the defines the bif: prefix
12:44:15
something along "sql: and bif: are reserved"
12:44:55
"400 Bad Request"
12:46:10
i get the same error on the dbpedia web ui, but not from arc
12:51:03
bengee thinks he found a work-around
12:51:35
you can define default namespaces in ARC via the config array
12:51:52
ah, cool
12:52:20
if you add "ns" => array("bif" => "http://example.com/"), the query will get through the parser, but it will be sent unchanged to the remote endpoint
12:53:29
it works, thanks!
12:53:33
cool
13:36:17
bengee: hugh williams replied to my quersion: http://boards.openlinksw.com/support/viewtopic.php?f=12&t=208
13:36:18
he asks:
13:36:20
In general with Virtuoso:
13:36:21
if, prefix bif: <nnn>
13:36:22
then, bif:contains will expand to nnn:contains
13:36:24
Question is does ARC2 perform the same expansion ?
13:36:27
---
13:36:45
i don't really how that can help with the problem
13:40:40
that doesn't look quite helpful...
13:40:49
let me see if I can get a better response.
13:41:02
arc-bot, pointer
13:51:56
looks like the query works if i drop the lang(?label) = "en"... although it took the best part of 600 seconds to complete :/
15:15:39
how many triples are in your store, arcnoob?
15:38:12
MacTed: you are iv_an_ru on the openlink forum?
15:38:55
ah, i just see iv_an_ru is someone else
15:39:23
but his answer in the forum also works:
15:39:24
If bif:contains is not handled without prefix declaration, write it as <bif:contains>, i.e. as an IRI, not as QName
15:42:50
ah, good
17:37:42
bengee - i've got around 1.5 million triples but saying that the server is pretty saturated in terms of cpu + memory... will try it on a beefer server in a couple of days.... thanks though for the quick turn around on adding maths to filter :) much apperciated :)
17:38:06
ok, welcome.
17:38:21
comparisons are not really optimized in arc
17:40:27
oh well on the optimizations... sure you'll get around to it at somepoint... at least its functional :)
17:40:44
go to dash but thanks for your help
17:40:56
your feedback will be helpful!