Next: Comparison of object models, Previous: The mini-oof.fs model, Up: Object-oriented Forth   [Contents][Index]

6.27.6 Mini-OOF2 ¶

Mini-OOF2 is very similar to Mini-OOF in many respects, but differs significantly in a few aspects. In particular, Mini-OOF2 has a current object variable, and uses the primitives >o and o> to manipulate that object stack. All method invocations and instance variable accesses refer to the current object.

>o ( c-addr – r:c-old ) new “to-o”

Set the current object to c_addr, the previous current object is pushed to the return stack

o> ( r:c-addr – ) new “o-restore”

Restore the previous current object from the return stack

To ease passing an object pointer to method invocation or instance variable accesses, the additional recognizer rec-moof2 is activated.

rec-moof2 ( c-addr u – nt translate-moof2 | translate-none  ) mini-oof2 “rec-moof-two”

This recognizer recognizes .word (if word is visible) and the translation is equivalent to >o word o>.

To assign methods to selectors, use xt class is selector, so no defines necessary. For early binding of methods, instead of :: you use defers:

[ class ] defers selector

Instead of writing

:noname code ; class is selector

you can also use the syntactic sugar

class :method selector code ;

:method ( class "name" –  ) gforth-experimental “colon-method”

define a noname that is assigned to the deferred word name in class at ;.