ooc-dev team mailing list archive

[Scott Olson <scott@xxxxxxxxxxxxxxx>]

You're right, we should use =name if the property is called `name`. 
Technically =arg should set the property named `arg`, regardless of which 
property's set() it's in.

On Monday 19 April 2010 07:21:39 Friedrich Weber wrote:
> Hi all,
> 
> I just added a basic implementation of the ORC in rock's `properties`
> branch. To test, use the `properties/001-dog.ooc` example in `tests`.
> 
> For now, supported features include getters and setters, accessing the
> actual value inside getters/setters, calling get/set inside
> getters/setters and replacing ordinary variable accesses `object
> property` and assignments `object property = value` with appropriate
> getter/setter calls.
> 
> Not working for now (but I'm working on this): automatic assignment
> arguments in setters (`set(=age)`), default getters/setters, missing
> getters/setters. I think this shouldn't take too long. ;-)
> 
> One implementation note: A property named `foo` creates a member `foo`
> in the struct, no name mangling is done. This should be okay, shouldn't it?
> 
> I noticed one thing in the ORC: We have this automatic assignment
> argument feature (`set(=age)`, aka AssArg). However, this would force
> the user to use the property name as argument name to avoid
> inconsistency with the regular automatic assignment argument, right?
> 
> Some code:
> 
> Dog: class {
> 	name: String {
> 		/* This would not work, right? */
> 		set(=arg) {
> 			/* ... */
> 		}
> 		/* rather, we'd need */
> 		set(=name) {
> 			/* ... */
> 		}
> 	}
> }
> 
> I think we should keep this behavior. Just wanted to make sure that this
> is wanted by everyone =)
> 
> Greetings,
> 
> Friedrich