Alzabo::RDBMSRules - Base class for Alzabo RDBMS rulesets

  use Alzabo::RDBMSRules;

  my $rules = Alzabo::RDBMSRules( rules => 'MySQL' );

This class is the base class for all CWAlzabo::RDBMSRules modules. To instantiate a subclass call this class's CWnew() method. See the SUBCLASSING Alzabo::RDBMSRules section for information on how to make a ruleset for the RDBMS of your choice.

A list of names representing the available CWAlzabo::RDBMSRules subclasses. Any one of these names would be appropriate as the rdbms parameter for the CWAlzabo::RDBMSRules->new() method.

The constructor always accepts one parameter, rdbms, which is the name of the RDBMS to be used.

Some subclasses may accept additional values.

The constructor returns a new CWAlzabo::RDBMSRules object of the appropriate subclass.

Throws: CWAlzabo::Exception::Eval Returns a list of SQL statements which would create the given schema. Returns a list of SQL statements to create the specified index. Returns a list of SQL statements to drop the specified table. Returns a list of SQL statements to drop the specified index.

This method takes two parameters:

This method compares the two schema objects and returns an array of SQL statements which turn the old schema into the new one.

This method takes two parameters:

This method compares the two table objects and returns an array of SQL statements which turn the old table into the new one. Returns a boolean indicating whether or not the column is numeric (integer or floating point).

Returns true or false to indicate whether or not the generated DDL SQL statements should have their identifiers quoted or not. This may be overridden by subclasses. It defaults to false.

If this is true, then when syncing a schema, the object will call CWalter_table_name_sql() to change the table's name. Otherwise it will call CWrecreate_table_sql().

If this is true, then when syncing a schema, the object will call CWalter_column_name_sql() to change the table's name. Otherwise it will call CWrecreate_table_sql().

The following methods are not implemented in the CWAlzabo::RDBMSRules class itself and must be implemented in its subclasses.

Returns a list of valid column types.

Given a string defining a feature, this method indicates whether or not the given RDBMS supports that feature. By default, this method always returns false unless overridden in the subclass.

Features that may be asked for:

* extended_column_types

Column types that must be input directly from a user, as opposed to being chosen from a list. MySQL's ENUM and SET types are examples of such types.

* index_column_prefixes

MySQL supports the notion of column prefixes in indexes, allowing you to index only a portion of a large text column.

* fulltext_indexes

This should be self-explanatory.

* functional_indexes

Indexes on functions, as supported by PostgreSQL. Throws an CWAlzabo::Exception::RDBMSRules if the schema's name is not valid. Throws an CWAlzabo::Exception::RDBMSRules if the table's name is not valid. Throws an CWAlzabo::Exception::RDBMSRules if the column's name is not valid.

Throws an CWAlzabo::Exception::RDBMSRules if the type is not valid.

This method returns a canonized version of the type. Throws an CWAlzabo::Exception::RDBMSRules if the length or precision is not valid for the given column.

This method takes two parameters:

This method is a bit different from the others in that it takes an existing column object and a potential attribute.

It throws an CWAlzabo::Exception::RDBMSRules if the attribute is is not valid for the column. Throws an CWAlzabo::Exception::RDBMSRules if the column is not a valid primary key for its table. Throws an CWAlzabo::Exception::RDBMSRules if the column cannot be sequenced. Throws an CWAlzabo::Exception::RDBMSRules if the index is not valid. Returns an array of SQL statements to create the specified table. Returns an array of SQL statements to create the specified column. Returns an array of SQL statements to create the specified foreign key. Returns an array of SQL statements to drop the specified column. Returns an array of SQL statements to drop the specified foreign key. Returns an array of SQL statements to add the specified column.

This method takes two parameters:

This method compares the two table objects and returns an array of SQL statements which turn the old table into the new one.

This method takes two parameters:

This method compares the two index objects and returns an array of SQL statements which turn the old index into the new one.

This method takes two parameters:

This method compares the two table objects and returns an array of SQL statements which alter the old one's primary key to match the new one's. Given a table, this method is expected to change the table's name from CW$table->former_name to CW$table->name. This will only be called if the rules object returns true for CWcan_alter_table_name(). Given a column, this method is expected to change the table's name from CW$column->former_name to CW$column->name. This will only be called if the rules object returns true for CWcan_alter_column_name().

This method takes two parameters:

This method is expected to drop the old table and create the new one.

However, it must preserve all the data stored in the old table, excluding data in columns that are being dropped. Additionally, if there are sequences associated with columns in the old table, they should not be dropped.

This method will only be called if either CWcan_alter_table_name() or CWcan_alter_column_name() return false. Given a schema object (which presumably has no tables), this method uses the schema's CWAlzabo::Driver object to connect to an existing database and reverse engineer it into the appropriate Alzabo objects. Returns a boolean indicating whether or not the column is an integer type. Returns a boolean indicating whether or not the column is a floating point type. Returns a boolean indicating whether or not the column is a character type. This is defined as any type which is defined to store text, regardless of length. Returns a boolean indicating whether or not the column is a date type. This is not true for datetime types. Returns a boolean indicating whether or not the column is a datetime type. This is not true for date types. Returns a boolean indicating whether or not the column is a time type. This is not true for datetime types. Returns a boolean indicating whether or not the column is a time interval type.

To create a subclass of CWAlzabo::RDBMSRules for your particular RDBMS is fairly simple.

Here's a sample header to the module using a fictional RDBMS called FooDB:

 package Alzabo::RDBMSRules::FooDB;

 use strict;
 use vars qw($VERSION);

 use Alzabo::RDBMSRules;

 use base qw(Alzabo::RDBMSRules);

The next step is to implement a CWnew() method and the methods listed under the section Virtual Methods. The new method should look a bit like this:

 1:  sub new
 2:  {
 3:      my $proto = shift;
 4:      my $class = ref $proto || $proto;
 5:      my %p = @_;
 6:
 7:      my $self = bless {}, $self;
 8:
 9:      return $self;
 10:  }

The hash CW%p contains any values passed to the CWAlzabo::RDBMSRules->new method by its caller.

Lines 1-7 should probably be copied verbatim into your own CWnew method. Line 5 can be deleted if you don't need to look at the parameters.

The rest of your module should simply implement the methods listed under the Virtual Methods section of this documentation.

Look at the included CWAlzabo::RDBMSRules subclasses for examples. Feel free to contact me for further help if you get stuck. Please tell me what database you're attempting to implement, and include the code you've written so far.

Dave Rolsky, <dave@urth.org>