This is a public chat log generated from the #semsol IRC channel.
09:41:13
bengee: started writing some notes: http://n2.talis.com/wiki/ARC2_Memory_usage
09:43:46
ah, cool
09:45:13
do you think some convenience method (such as parser->reset()) could make sense here?
09:47:05
I could also think about not setting the dataype and lang if they are not present
09:47:19
for a lower per-triple footprint
09:47:39
although we might need more isset() checks in other places then
09:50:43
bengee: sounds good - an additonal benefit would be enabling reuse of the parser for different docs?
09:51:45
hmm, perhaps
11:00:49
bengee: yep, a parser->reset() to clear it would be nice
11:01:12
bengee: as would not setting the datatype and lang, but not too worried about those
11:01:53
bengee: on a more complex note, you could abbreviate the URIs with namespaces, but that would make lots of the graph handling have to change
11:02:40
yep, and don't let talis hear that ;)
11:03:20
mmmmmrob-- ;)
11:03:38
bengee likes the current rdf/json rdf/php approach
11:03:56
phew :)
11:06:06
kwijibo: eh? what did I get a karma hit for?
11:06:11
mmmmmrob is trying to help
11:06:32
;) suggesting namespace abbreviations
11:06:38
ah
11:06:54
It was just a thought, that's why I added that it wuld be copmplex
11:07:16
thus undoing months of discussion and code alignment ;)
11:07:38
kwijibo: why would you not want to though? other than it makes things harder - for the internal arrays only, not for the simple Index or anything
11:08:31
I wouldn't want to if it were my code - I'd just say 'buy more memory'
11:08:42
but that's cause I'm lazy and like simpler code
11:08:46
spose - though what you save on memory you'd maybe lose on performance
11:08:53
true
11:09:22
you would have to do a lot of string manipulation inbound and outbound
11:09:29
probably kills it
11:10:08
i wonder if you could do anything with constants or something
11:10:19
anyhoo, now we know the memory usage is linear and ~2k per triple and that we can discard the parser (helper method would be great) it's not too much of a concern
11:11:07
i guess the constant would just get converted to a string on assignment, i'm talking rubbish
11:12:51
I've just added a reset() method to the RDF parsers, and tweaked the getSimpleIndex builder to use in_array + an array as needle for dupe checks
11:13:36
thanks a lot for exporing this, mmmmmrob
11:13:42
exploring even
11:14:18
bengee: np, we're using it, so would like to contribute back if we can
11:14:48
maybe I can make up for the all mess^H^H^H^H help kwijibo's been giving you
11:15:14
mmmmmrob ducks
11:15:35
:D
11:21:58
;)