Refactor how column families and it's fields are maintained in ORM table class about massive_record HOT 2 CLOSED

Progress has been made on this, see comparison between develop and my refactor branch: develop...feature/refactor_schema

I think these changes are good, and in short we now have:

• Made ColumnFamilies into a set; it is no longer possible to have column families with the same name.
• Made Fields into a set; it is no longer possible to have fields with the same name.
• The enforcement of fields with unique names are global across all column families.
• The attributes_schema() is read from from the column family set, and there are no longer a need to maintain multiple hashes which reflected the column families and fields in different ways.
• Added simple interface for dynamically adding fields to a family: ATableClass.add_field_to_column_family(:info, :name, :default => "NA"). It will take care of keeping the ATableClass consistent with read/write and other attribute methods it creates based on the schema.
• ORM::Column is now implemented, mening that it should be easier to subclass it, and define fields given the DSL we have defined in the README.
• ORM::Column has attributes_schema(), known_attribute_names(), default_attributes_from_schmea() just like the ORM::Table. In fact, these methods comes from the same module.
• All old tests passes, and I suspect that we should not need to do a lot of changing in the Front. What we can do in the front is clean up classes which inherits from ORM::Column, as we can now define fields for these classes as well :-)

from massive_record.

Merged into master :-)

from massive_record.