NAME¶
SQL::Translator::Diff - determine differences between two schemas
DESCRIPTION¶
Takes two input SQL::Translator::Schemas (or SQL files) and produces ALTER
statments to make them the same
SNYOPSIS¶
Simplest usage:
use SQL::Translator::Diff;
my $sql = SQL::Translator::Diff::schema_diff($source_schema, 'MySQL', $target_schema, 'MySQL', $options_hash)
OO usage:
use SQL::Translator::Diff;
my $diff = SQL::Translator::Diff->new({
output_db => 'MySQL',
source_schema => $source_schema,
target_schema => $target_schema,
%$options_hash,
})->compute_differences->produce_diff_sql;
OPTIONS¶
- ignore_index_names
- Match indexes based on types and fields, ignoring
name.
- ignore_constraint_names
- Match constrains based on types, fields and tables,
ignoring name.
- output_db
- Which producer to use to produce the output.
- case_insensitive
- Ignore case of table, field, index and constraint names
when comparing
- no_batch_alters
- Produce each alter as a distinct "ALTER TABLE"
statement even if the producer supports the ability to do all alters for a
table as one statement.
- ignore_missing_methods
- If the diff would need a method that is missing from the
producer, just emit a comment showing the method is missing, rather than
dieing with an error
PRODUCER FUNCTIONS¶
The following producer functions should be implemented for completeness. If any
of them are needed for a given diff, but not found, an error will be thrown.
- •
- "alter_create_constraint($con)"
- •
- "alter_drop_constraint($con)"
- •
- "alter_create_index($idx)"
- •
- "alter_drop_index($idx)"
- •
- "add_field($fld)"
- •
- "alter_field($old_fld, $new_fld)"
- •
- "rename_field($old_fld, $new_fld)"
- •
- "drop_field($fld)"
- •
- "alter_table($table)"
- •
- "drop_table($table)"
- •
- "rename_table($old_table, $new_table)"
(optional)
- •
- "batch_alter_table($table, $hash)" (optional)
If the producer supports "batch_alter_table", it will be called
with the table to alter and a hash, the keys of which will be the method
names listed above; values will be arrays of fields or constraints to
operate on. In the case of the field functions that take two arguments
this will appear as a hash.
I.e. the hash might look something like the following:
{
alter_create_constraint => [ $constraint1, $constraint2 ],
add_field => [ $field ],
alter_field => [ [$old_field, $new_field] ]
}
- •
- "preprocess_schema($class, $schema)" (optional)
"preprocess_schema" is called by the Diff code to allow the
producer to normalize any data it needs to first. For example, the MySQL
producer uses this method to ensure that FK contraint names are unique.
Basicaly any changes that need to be made to produce the SQL file for the
schema should be done here, so that a diff between a parsed SQL file and
(say) a parsed DBIx::Class::Schema object will be sane.
(As an aside, DBIx::Class, for instance, uses the presence of a
"preprocess_schema" function on the producer to know that it can
diff between the previous SQL file and its own internal representation.
Without this method on th producer it will diff the two SQL files which is
slower, but known to work better on old-style producers.)
AUTHOR¶
Original Author(s) unknown.
Refactor/re-write and more comprehensive tests by Ash Berlin
"ash@cpan.org".
Redevelopment sponsored by Takkle Inc.