This is a public chat log generated from the #semsol IRC channel.
09:05:55
I have a basic question facing the components of a semantic web application
09:06:17
let's say i want to build a foaf editor
09:06:21
what would you do:
09:07:41
1) store data in mysql, edit data over REST/XML-RPC API, publish data by constructing rdf in a php script, then loading this RDF-data via the ARC2 Triple Store, working with the data via ARC2
09:08:11
2) store data directly in the triple store, edit data over SPARQL Endpoint, publish data simply via ARC2
09:08:36
is it save to store the data directly and only in the triple store?
09:09:01
what experience did you make?
09:17:36
I personally do everything in RDF/SPARQL these days
09:18:25
I sometimes use multiple stores, though
09:19:17
i.e. you work with small "editing" stores, and less restricted "publish" stores
09:20:06
if you expect a lot of traffic, it might also make sense to store files and to just load them into a sparql store for editing
09:21:14
there are some implementations where the rdf is edited by entirely separate tools and just loaded into an rdf store for aggregated querying
09:25:41
I guess two decision factors are: how much data are you going to work with (because Updates get slow in larger RDF stores), and how familiar you are with SPARQL (setup vs. coding efficiency)
09:32:03
ok thx, this helps, maybe i'll try it with two stores then, one for my personal foaf file and one for loading many other foaf profiles with little inferencing
09:32:56
yeah, sounds good
09:33:34
I also tried to use the sparql endpoint for editing a foaf profile directly in a adobe flex client
09:33:48
but this wasn't very successful
09:34:37
oh, anything arc-wise that didn't work?
09:34:41
because the whole application is now very event based, and e.g. if I want to edit a triple I have to delete it and then insert it, but this isn't very safe, because the connection could abort in between those two queries
09:34:58
ah, I see
09:35:12
now I will switch to a REST-API which can do those things then on the server
09:35:45
also using arc directly in php (and not over the endpoint) is much more powerful
09:42:45
yes, I'm doing it like that, too. I'm using a little framework to handle the requests, with lots of local sparql queries