You can save Transformer models in two formats: .py? and .mdl. You can only open models saved as .py? files in the Transformer (Windows) interface. However, you can open models saved as .mdl files in either the Windows interface or a text editor.
Each .mdl file is a plain-text representation of the model, and may be compatible with other versions of Transformer than the one used to create it. However, it loads more slowly than a .py?-format file because it recreates all of the Transformer objects each time it loads.
You can open models saved as .mdl files on any supported UNIX, Linux, or Windows platform, making this the preferred format to use in a mixed production environment.
The .py? format is a binary representation of the model. Although not compatible between versions of Transformer, it loads faster than the .mdl-format equivalent, even though it is larger, because the Transformer objects are not recreated when the model reloads.
As you edit a model, the size of the associated .py? file increases because information about the operations performed during model editing is retained in the file. Transformer uses this information when performing incremental updates and other subsequent cube operations. Lengthy descriptions of the appended operations can cause fragmentation of the binary file. We therefore recommend that you periodically save your .py? models as .mdl files.
