"Adam Megacz" <[EMAIL PROTECTED]> wrote in message news:[EMAIL PROTECTED] > "Igor Tandetnik" <[EMAIL PROTECTED]> writes: >>> I have an application that absolutely must not return from a certain >>> call until the results of an update are safely committed to disk. > >> Open a separate connection to the same database, perform the update >> on this connection. > > I am in an environment where the underlying operating system does not > provide reliable interprocess file locking (in fact, it flat-out > lies). Is this still safe?
Wouldn't you have two connections within the same process, in your scenario? As I understand, within the same process SQLite doesn't rely on POSIX locks and instead uses in-memory data structures shared by all connections. There was a similar discussion recently - see http://thread.gmane.org/gmane.comp.db.sqlite.general/35784/ Igor Tandetnik _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
