I know the restriction of passing open databases across threads. What I'm wondering is whether that's really a process/thread issue, or a concern over making sure two or more threads don't muck with a single database at the same time.
Basically, I'm exploring how I might implement database access in a Web server environment. One way is to have a single thread manage all the database access and have the individual user threads queue up requests to that db thread. If really, the restriction is one access at a time, another way may be to have a semaphore that determines if the database is being accessed and, if not, any given thread could set the busy semaphore and then access the database. Obviously, there are some race condition considerations, but is this approach possible as well? Is it really "just don't access the database simultaneously" or is the concern really specific to the thread mechanism? Thanks! -- David ----------------------------------------------------------------------------- To unsubscribe, send email to [EMAIL PROTECTED] -----------------------------------------------------------------------------
