in reply to Parallel::ForkManager and BEGIN/END oddity...?

Why not simply use "LOCK TABLES" and "UNLOCK TABLES" and let the SQL server manage locks ?

  • Comment on Re: Parallel::ForkManager and BEGIN/END oddity...?

Replies are listed 'Best First'.
Re^2: Parallel::ForkManager and BEGIN/END oddity...?
by atcroft (Abbot) on Oct 27, 2003 at 11:22 UTC

    Because (as I mentioned in my second paragraph, above):

    Because the version of the database I am having to deal with uses table-level locking, I figured that semaphores might be a way to let the children know when one or another was using the tables.
    If I am understanding what (UN)LOCK TABLES does, it won't help, since the very problem I am trying to work around is "dead time" in the child processes where the table is locked by one process and thus unwritable by other children.

    Apparently, as rewritten, the PID file now contains just a number of lines equal to one plus the number of child runs.

    Any additional feedback would be greatly appreciated.