summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
-rw-r--r--FS/FS/cust_pkg.pm10
1 files changed, 7 insertions, 3 deletions
diff --git a/FS/FS/cust_pkg.pm b/FS/FS/cust_pkg.pm
index dff241393..d2d74190d 100644
--- a/FS/FS/cust_pkg.pm
+++ b/FS/FS/cust_pkg.pm
@@ -107,11 +107,12 @@ sub insert {
# custnum might not have have been defined in sub check (for one-shot new
# customers), so check it here instead
+ # (is this still necessary with transactions?)
my $error = $self->ut_number('custnum');
- return $error if $error
+ return $error if $error;
- return "Unknown customer"
+ return "Unknown customer ". $self->custnum
unless qsearchs( 'cust_main', { 'custnum' => $self->custnum } );
$self->SUPER::insert;
@@ -567,7 +568,7 @@ sub order {
=head1 VERSION
-$Id: cust_pkg.pm,v 1.7 2001-10-01 10:31:08 ivan Exp $
+$Id: cust_pkg.pm,v 1.8 2001-10-09 03:11:50 ivan Exp $
=head1 BUGS
@@ -584,6 +585,9 @@ cancel } because they use %FS::UID::callback to load configuration values.
Probably need a subroutine which decides what to do based on whether or not
we've fetched the user yet, rather than a hash. See FS::UID and the TODO.
+Now that things are transactional should the check in the insert method be
+moved to check ?
+
=head1 SEE ALSO
L<FS::Record>, L<FS::cust_main>, L<FS::part_pkg>, L<FS::cust_svc>