We have sqlite databases in a memory shared cache environment where individual tables may be locked out by other processes. This means that we have to worry about SQLITE-BUSY errors and make sure are code can handle this.
There is some internal debate about whether we have to worry about table locks only at the beginning of a transaction, or whether a commit itself can generate them. The question is, if we have a transaction with several sql statements that execute without any locks, is there anyway that the commit itself would encounter a lock. We know that the commit involves writing of the results from memory to disk, but would any potentials locks actually be deferred until this point? _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users