Am 19.04.2006 um 15:31 schrieb [EMAIL PROTECTED]:
Felix Schwarz <[EMAIL PROTECTED]> wrote:
Hi,
I have just discovered these old message on the SQLite Yahoo! Group:
http://groups.yahoo.com/group/sqlite/message/2115
http://groups.yahoo.com/group/sqlite/message/2117
This does sound very interesting! If I simply want to store binary
data that I want to access via an index number again (and do this
real quick), how big is the difference between using SQLite or its
btree functions directly?
I'm guessing that you will not see a significant performance
improvement over using prepared statements in the SQL interface
and a table like this:
CREATE TABLE storage(
id INTEGER PRIMARY KEY,
data BLOB
);
Thank you very much. Regarding the structure, that's exactly what I
had in mind.
I've now started to extend my SQL wrapper classes to support caching
of prepared statements. This has, however, raised a question for me
regarding an edge case:
I might have to call sqlite3_finalize() on a thread different from
the one the initial sqlite3_prepare() call took place. My classes
take care that there's one sqlite3-connection per thread, so the
sqlite3-connection used with the initial sqlite3_prepare() call has
also been opened on that other thread.
Example:
Thread A opens sqlite3-connection connA
Thread A prepares a statement stA using sqlite3_prepare() using the
connection connA
Thread B calls sqlite3_finalize on statement stA
Is this a safe operation to do?
Thanks again.
Felix
