[redland-dev] duplicate bnode ids

Dave Beckett dave.beckett at bristol.ac.uk
Fri Sep 17 10:42:54 BST 2004


A possible answer to getting dups.

The way that redland makes bnodes is to take the startup time of
the application and add an incrementing ID suffix.  So... I think
the problem here is multiple invocations of rdfproc.  The fix
is to add 'sleep 1' between calls.

Yeah this is a hack, but it usually works since mostly you don't
start redland many times a second.  This is done since at present there
is no extra outside-database state such as IDs, recorded in redland
persistent stores.  It's possible I'll add this as it is handy to
record store metadata.

To fix this particular problem I might add gettimeofday() in secs
plus some digits from microseconds, just makes the IDs a little longer :)

Dave




More information about the redland-dev mailing list