This is a public chat log generated from the #semsol IRC channel.
23:05:18
doh, can just parse and reserialize 3 times and check that 2 and 3 are the same - that should be ok
23:05:31
and they are, whoohoo
23:19:02
http://n2.talis.com/svn/playground/kwijibo/PHP/arc/plugins/trunk/ARC2_SPARQLSerializerPlugin
08:06:49
bengee: I am wondering about dpeendencies between plugins - just document them, or include them in the same folder and require them (instead of using the ARC:inc )
08:07:30
hmm
08:07:49
document + ARC::inc perhaps?
08:09:00
okiley dokiley ;)
08:09:28
you could perhaps print an error if ARC::inc returns false
08:10:09
your plugin is already on /download/plugins, btw
08:10:22
cool
08:11:48
ooh, therre's also this one http://n2.talis.com/svn/playground/kwijibo/PHP/arc/plugins/trunk/ARC2_IndexUtils/
08:11:59
*oh
08:12:53
."Plugin", and I'll add it
08:12:59
;)
08:13:05
oh yeah :)
08:14:59
I was thinking last nigth that it would be cool to hack the sparqlparser to be more readily extensible, by passing in additional allowed query types and functions etc
08:15:09
I might have a hack on that later on
08:15:16
cool
08:15:51
the use case would be for stuff like allowing ABOUT { ?s ?p ?o }
08:15:57
catching that
08:17:00
and rewriting it to CONSTRUCT { ?stmnt rdf:subject ?s ; rdf:predicate ?p ; rdf:object ?o } WHERE etc
08:17:19
sorry, etc not obvious :p
08:17:25
WHERE { ?stmnt rdf:subject ?s ; rdf:predicate ?p ; rdf:object ?o }
08:18:04
I dunno, just making it easier to experiment with stuff really
08:18:39
would be cool if there was a way to hook into ARC::inc, so that you could specify e.g. $conf = array('class_replacements' => array('SPARQLParser' => 'MySPARQLParser)) which would then be checked when inc() is called
08:19:45
yeah, I suppose you might want to do the same for, eg, Remote SparqlEndpoint, or the TalistStorePlugin
08:20:06
although not so much maybe
08:21:25
maybe the SPARQL thing is not so much a straight replacement - you also want another layer that does the rewriting
08:22:01
hmm, triggers?
08:22:59
are triggers just for after an operation?
08:23:31
after queries
08:24:55
is there any way to set a variable in a non-instantiated class?
08:25:11
not that I can think of
08:25:23
self::foo = ...?
08:25:24
you mean in a statically called class?
08:31:21
bengee what about a static variable within the method ?
08:31:34
yeah, just investigating that
08:31:56
you could check if it isset, then set it from within the mtheod, or something like that
08:33:15
hmm "it's impossible to assign a variable result to a static variable"
08:34:55
that's a bit useless!
08:37:20
I killed the load script last night btw, it had got to 642000 triples
08:37:32
not that that's very useful information since it was before the patch
08:37:59
'k
08:49:12
hmm, maybe that was a mis-info, you just can't assign a variable value via static $my_var = $val, but you can do static $my_var = ''; $my_var = $val
08:51:53
yep, worky
08:53:07
function setStatic($val) {
08:53:09
static $arc_static = '';
08:53:11
if ($val) $arc_static = $val;
08:53:13
if (!$val) return $arc_static;
08:53:14
}
08:53:17
function getStatic() {
08:53:19
return ARC2::setStatic('');
08:53:21
}
08:53:34
bengee tries it with php5
08:54:24
seems to work, too
08:56:23
cool
08:58:01
now I could implement a ARC::replaceClass($old_class_name, $new_class_name) and link those defs with ARC::inc()
08:59:54
coolio
09:01:42
so the other part of the puzzle, I guess, would be before-triggers, in order to have the architecture amenable to SPARQL rewriting for stuff like danbri's ACL
09:02:20
not sure how necessary a feature it is though, so I won't push you for it ;)
09:02:32
you could simply sub-class/replace the sparql parser
09:02:47
yes
09:02:54
like you do for the sparqlplusparser
09:11:10
hm, the Class ... extends Parent will still point at the old parent, though
09:13:50
so this could only work for top-level classes
09:15:10
err, bottom-level, whatever
09:46:50
hmm, maybe it would be better to import/export in chunks of triples, instead of resources
11:15:43
hey bengee!
11:16:00
heya
11:16:04
what's the best way to cancel a query when it's taking too long
11:16:23
best way means less destructive!
11:16:37
killing the mysql process
11:17:03
well, isn't there a more gentle way. I imagined rebooting mysql for ex
11:17:19
I'm trying to automatize that
11:17:27
rebooting will cancel all running processes
11:17:39
and I don't want to end up killing processes if possible
11:18:12
well, it's in a benchmark context so I don't mind.
11:18:21
but I understand that on a production context it's different
11:18:42
so there is no other way to do that through ARC2 then
11:19:03
say setting up a timeout to a query before you send it for example
11:19:10
not yet. there are hooks in the endpoint class, but the code isn't in trunk yet
11:19:30
alright
11:19:55
I'm using SHOW PROCESSSLIST, and then kill all process where "time" is more than the specified timeout
11:20:31
pretty simple to implement, but not added to ARC2 yet
11:23:55
bengee: +1
11:23:58
:)