On 07/02/12 11:33, Milorad Tosic wrote:
Hi Emillio,
Two small follow-up questions:
1) Having two different APIs for query and update SPARQL means that
in a SPARQL endpoint we should first try the query and if it throws
exception then we try update API? Or there is something else to deal
with the separation between the two APIs in processing a general
SPARQL request?
You can parse the request with one API then the other to see what sort
of request it is but I would point out that update operations are rather
different to query especially as to who may do them. It's been a
weakness of SQL (IMO) that everything is one language and goes over one
JDBC connection.
2) Speaking about synchronization, does it mean that query model is
thread safe and we do not have to concern concurrent query requests?
We only need to do that for update?
Locally: multiple reader (query) OR single writer (update) - not both.
Remote (Fuseki): request are executed in a safe manner (previously by
locking, now transactions for TDB).
Andy
Thanks, Milorad
