Fix non-deterministic UTF8 handling with PostgreSQL database

If using PostgreSQL and versions of DBD::Pg >= 3, code will behave differently if no value is set in config for DatabaseDefault PG_ENABLE_UTF8. This would affect installations which were even opting out of all UTF-8 support in Interchange by using the MINIVEND_DISABLE_UTF8 environment variable. This happened because the default UTF-8 handling for DBD::Pg as of 3.0 turned into automatically decoding UTF-8 data as perl scalarsm so catalogs/stacks which were now using a new version of DBD::Pg could run into unexpected behavior (at the very least, increased parsing time due to using unicode-aware regexes under-the-hook). Specify the default to be off to match the existing state before this module; in all cases, you have to opt in to UTF-8 specific IC, so no reason to change this here. Reviewed-by: Mark Johnson <mark@endpoint.com>
interchange · Nov 14, 2017 · 459ffdb · 459ffdb
1 parent b6ebec3
commit 459ffdb
Showing 1 changed file with 2 additions and 0 deletions.
diff --git a/lib/Vend/Config.pm b/lib/Vend/Config.pm
@@ -4526,6 +4526,8 @@ sub parse_database {
 		else 						{ $d->{Class} ||= $Global::Default_database	}
 
 		if($C and $C->{DatabaseDefault}) {
+			$C->{DatabaseDefault}{PG_ENABLE_UTF8} = 0
+				unless defined $C->{DatabaseDefault}{PG_ENABLE_UTF8};
 			while ( my($k, $v) = each %{$C->{DatabaseDefault}}) {
 				$d->{$k} = $v;
 			}