Short version: Does anyone use with either of the
rdflib mysql, postgresql or sqlite stores? Are you happy?
Long version:
RDFLib has a range of stores for persistent graphs in a RDBMS:
rdflib-mysql, rdflib-postgresql, rdflib-sqlite ...
These all build on a single AbstractSQLStore - currently living in RDFExtras.
We've always migrated these stores along, and run the tests once in
while. Unfortunately I have the feeling they are not really
maintained, and that all have pretty serious issues that we do not
know about (known unknown, etc.)
* They've seen almost no code changes in the last 2-3 years. (Since
the rdflib-layercake split-off - I know Chimezie has done some
maintenance on the mysql side of things in layercake)
* We receive very few support request - the few ones we get are often
of the type "this is completely broken"
* The documentation is outdated.
* Finally, I did not ever personally used them :)
Now - does anyone really use these? Do they function in your "daily
work" with rdflib?
I do not personally have the resources to maintain all of these, the
AbstractSQL store is rather complicated, and I would have to spend a
lot of effort to understand it before being able to fix anything.
Also, the main benefit of AbstractSQL store was the old Sparql2SQL
code that would convert most of sparql 1.0 code to SQL queries,
allowing SPARQL eval on the server-side. We did however remove this
code from rdfextras some years back (it was too complicated for any of
the remaining rdflib developers to maintain - again, I know Chimezie
has kept this code and done some maintenance in rdflib-layercake)
If noone steps up to tell me they work just fine - or even better,
steps up to do some maintenance, I would like to mark each of them
with a big-fat "ABANDONED CODE - NO LONGER MAINTAINED" warning, and
perhaps focus on a SQLAlchemy backed store?
Any thoughts?
- Gunnar
--
http://gromgull.net