|Think about Loose Coupling|
Re: DBI, fork, and clone.by mr_mischief (Monsignor)
|on Jan 16, 2007 at 01:11 UTC||Need Help??|
Great work. I don't even get the warning you mention on my test box.
Note that DBI's clone() method is labelled as new and subject to change. Also worth noting is that the username argument to DBI's connect is just that -- 'Username', to be exact.
I mentioned that I don't get the warning you noted in your node. I was thinking that things sounded odd considering the 'User'/'Username' business, so I tried changing the $dbh->clone() call like this:
... and I got exactly the warning you mentioned!
My guess is that your particular DBI (or DBI/DBD combination) is connect()'ing with a 'Username' attribute and then re-issuing it with something additional called 'User'. I'm using MySQL, and it looks like you're using Postgres from the driver info you had in the test. I'll give you my info for comparison:
The perl binary I tested against does have Mandriva-applied patches, but I doubt that has anything to do with the matter.
ActiveState perl 5.8.7 on Windows with DBI 1.4.2 and DBD::mysql 3.0002 fails to create a clone, BTW. It gives the following complaint: "DBD::mysql::db clone failed: handle 2 is owned by thread 2244a4 not current thread 1cccad4 (handles can't be shared between threads and your driver may need a CLONE method added) a t db_clone line 59". Probably has to do with Windows not supporting fork() and the perl installation having to make do with threads, smoke, and mirrors.
Christopher E. Stith