What is the difference between $DBI::errstr and $dbh->errstr?

[BBQ]

The question says it all. While reading through perltoot, I noticed that the both look alot like (respectivly) a function call and a method call. But what of the practical use for either of them? In what do they differ?

Originally posted as a Categorized Question.

[btrott]

$DBI::errstr is actually a variable, not a method call. It holds the text of the last error on any of the database handles.

$dbh->errstr is a method call on a database handle. It returns the text of the last error from "the last driver function called," according to the DBI manpage.

In general, you should use $dbh->errstr, unless you don't have an available database handle. For example, if you're calling DBI->connect, you're creating a database handle, so you don't actually have one to call the errstr method on. So if you're doing error checking on the connect method, you should print out $DBI::errstr if you get undefined back from connect.

When you're doing error checking for any database handle methods (prepare, execute, etc.), you should use $dbh->errstr. (Or just set RaiseError.)

[archfool]

As a sidenote, if you are using PL/SQL in DBD::Oracle, there is a special method for PL/SQL errors:

        my $msg = $dbh->func( 'plsql_errstr' );

        die $dbh->errstr if ! defined $msg;
        die $msg if $msg;