On Tue Aug 13 08:48:52 2013, BRAD wrote:
Show quoted text> Hi, having problems with DBIx::Class::DeploymentHandler -
> SQL::Translator::Diff bombs out when generating Oracle
> upgrade/downgrade scripts with the error in the title. There isn't a
> drop_table method in SQL::Translator::Producer::Oracle, so I hacked
> the following from SQL::Translator::Producer::MySQL:
>
> 492 sub quote_table_name {
> 493 my ($table_name, $qt) = @_;
> 494
> 495 $table_name =~ s/\./$qt.$qt/g;
> 496
> 497 return "$qt$table_name$qt";
> 498 }
> 499
> 500 sub drop_table {
> 501 my ($table, $options) = @_;
> 502
> 503 my $qt = $options->{quote_table_names} || '';
> 504
> 505 my $table_name = quote_table_name($table, $qt);
> 506 return (qq[DROP TABLE $table_name CASCADE CONSTRAINTS]);
> 507 }
>
> I'm happy to submit a proper patch for this (and the missing
> drop_field method) later on if needed.
Patch to add the three methods:
--- /home/a530093/perl5/lib/perl5/SQL/Translator/Producer/Oracle.pm.orig 2013-08-21 16:03:55.910188883 +0100
+++ /home/a530093/perl5/lib/perl5/SQL/Translator/Producer/Oracle.pm 2013-08-21 16:07:04.199195125 +0100
@@ -489,6 +489,37 @@
return \@create, \@fk_defs, \@trigger_defs, \@index_defs, ($options->{delay_constraints} ? \@constraint_defs : []);
}
+sub quote_table_name {
+ my ($table_name, $qt) = @_;
+
+ $table_name =~ s/\./$qt.$qt/g;
+
+ return "$qt$table_name$qt";
+}
+
+sub drop_table {
+ my ($table, $options) = @_;
+ my $qt = $options->{quote_table_names} || '';
+ my $table_name = quote_table_name($table, $qt);
+ return (qq[DROP TABLE $table_name CASCADE CONSTRAINTS]);
+}
+
+sub drop_field
+{
+ my ($old_field, $options) = @_;
+
+ my $qf = $options->{quote_field_names} || '';
+ my $qt = $options->{quote_table_names} || '';
+ my $table_name = quote_table_name($old_field->table->name, $qt);
+
+ my $out = sprintf('ALTER TABLE %s DROP COLUMN %s',
+ $table_name,
+ $qf . $old_field->name . $qf);
+
+ return $out;
+
+}
+
sub alter_field {
my ($from_field, $to_field, $options) = @_;
Tested on a fairly small codebase, the generated schema alterations look OK however I've not had a chance to test them in Oracle.
Cheers
Brad
