🚀 go-pugleaf

In article <4kfra6$k1t@tuegate.tue.nl>,
MWeiss  <MartinW@ctrl.phys.tue.nl> wrote:
>MODULE Test;
>
....
>	
>PROCEDURE Do*;
>	VAR x: T1; next: T0;
>BEGIN
>	NEW(x);
>	x :=  x.next(T1); (* a TRAP is generated here *)
>END Do;
>
>END Test.
>
>It seems that type-guarding the pointer x.next which happens to 
>have the value NIL at execution time causes the problem. Did 
>anybody else encounter this problem before? 

Yes. NEW (x) obviously initializes x.next to NIL. You cannot type-guard a
pointer whose value is NIL. Therefore the trap.
You must always ensure that the pointer being guarded is not NIL before
guarding it, e.g. IF x.next # NIL THEN x := x.next(T1) END
>
>I am using Oberon/F v 1.1 for Windows.

You'll find this behaviour in all (?) Oberon implementations

-Stephan

------------------------------------------------------------------------------
Stephan Gehring                            Email:  gehring@inf.ethz.ch
Institute for Computer Systems             Phone:  +41 1 632 73 15
ETH-Zentrum                                FAX:    +41 1 632 13 07
CH-8092 Zurich, Switzerland