On 5/9/07, Chas Owens <[EMAIL PROTECTED]> wrote:
On 5/9/07, Jeff Pang <[EMAIL PROTECTED]> wrote:
snip
> Yes but I think is this maybe a bug in DBI class?For example,open a file
handle and
> after forking child close that handle,this would not affect parent's handle.
snip
It definitely violates the principle of least surprise, but it is also
a very hard problem. File handles are much simpler* in comparison.
The handle is auto-disconnecting so that you don't accidentally exit
your script leaving the database connection open. Imagine a script
that connects to the database, runs a short query, and then ends. Now
imagine that script running several times a second. Each time it runs
it leaks a connection that is not reclaimed by the server until it
times out (if it times out). The proper way to handle this (in my
opinion, which might be terribly flawed since I haven't thought about
it very hard yet) is to have some sort of flag that can be set on the
database handle to tell the DBI to destroy it without closing the
connection. This flag could be set in the child and everybody would
be happy. I am looking in the docs now to see if such a flag exists.
If it doesn't then I am going to bug the maintainers of both the DBI
and DBD::mysql to add it.
And happily such a flag exists: InactiveDestroy.
Just add
$dbh->{InactiveDestroy} = 1;
to the child and the problem should go away.
--
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
http://learn.perl.org/
