nathankramer / apex-rumble Goto Github PK

Consider some smart way of avoiding real IO to the file system when testing the build functionality

RumbleModel
├─ RumbleAccount (generated from Account)
│  └─ MAccount
├─ RumbleAsset (generated from Asset)
|  └─ MAsset
├─ RumbleContact (generated from Contact)
|  └─ MContact
└─ RumbleProject (generated from Project__c)
|  └─ MProject

I've proposed a sort of three tiered structure.
At the root we have the RumbleModel base class.
Next we have the generated model classes, and finally at the bottom we have the actual "Model" class that the developer uses.

Is the class naming OK?

The M prefix is my biggest concern, but I like it for a couple of reasons:

• Clear differentiation from the volatile / generated classes
• Nice and short
• Still has the benefit of grouping in the text editor without folder structure as an option

Original plan of assuming the user has a way of retrieving metadata, eg mavensmate, and parsing the present src/objects/ folder is going to give us issues when standard objects are involved.

Other ideas we've considered are a configuration file that lists the desired objects.
We could still allow for parsing objects that exist and then building up this config file?