RE: [Gm2] equality, inequality tests and structured types

[Breeden, Thomas (tmb)]

> -----Original Message-----
> From: Andreas Fischlin [mailto:address@hidden
> 
> We need to keep these issues separate, i.e. equality and inequality
> tests and coercability between pointers (ADDRESS) and variables of an
> opaque type. Therefore:
> ...
>
> 
>      IF x1 <> x2 THEN
> 
> has not necessary anything to do with assignment compatibility. A
> construct like
> 
>           undefMyOpaque: MyOpaque; (read-only)
> 
> can serve to overcome this NIL incompatibility 

Ah yes. I have gotten so unaccustomed to putting variables in the Definition 
module,
I have been doing this with a function call like IOChan.InvalidChan() to be 
safe.

Tom
address@hidden