Bug with 1.2.18 on MacOS

[Jasper Taylor]

Hi folks...I just built GNU prolog on my new Mac Mini using the file 
gprolog-20040608.tgz from the 'unstable' directory, and got some very strange 
behaviour. It seems variables are keeping their values from an operation even 
after it fails. Here is an example:
---------------------------------------------------------------
GNU Prolog 1.2.18
By Daniel Diaz
Copyright (C) 1999-2004 Daniel Diaz
| ?- trace.
The debugger will first creep -- showing everything (trace)

yes
{trace}
| ?- foo(bar) =.. [A, B, C]; A=B.
       1    1  Call: foo(bar)=..[_18,_19,_20] ?
       1    1  Fail: foo(bar)=..[foo,bar,_20] ?
       1    1  Call: foo=bar ?
       1    1  Fail: foo=bar ?

no
{trace}
| ?-
-----------------------------------------------------------------
This fragment seems to work OK on Windows and Linux, what could be going on? 
Also I think the stable 1.2.16 is OK on MacOS too.

Cheers
 --Jasper