Am Dienstag, den 01.12.2009, 01:03 -0700 schrieb Robert Dodier:
> On 11/30/09, Dieter Kaiser <drdieterkaiser at web.de> wrote:
>
> > There are two changes in the output visible to the user:
> >
> > 1. depends now add the new dependencies always at the end of the
> > property list.
> > 2. The return value of depends shows the complete dependencies,
> > including the dependencies already on the property list.
> >
> > (%i4) depends(f,[x,y]);
> > (%o4) [f(x,y)]
> > (%i5) depends(f,[a,b]);
> > (%o5) [f(x,y,a,b)]
>
> Hmm. I'm not in favor of this. I'm pretty sure it is clearer to have
> the effect of a declaration supersede previous declarations.
> Otherwise to know the effect of a new declaration, you have to
> know any previous ones.
Yes, you are right. I have thought about it too and the code would
become a bit easier, but that is the current behavior:
(%i1) depends(f,x);
(%o1) [f(x)]
(%i2) depends(f,y);
(%o2) [f(y)]
(%i3) dependencies;
(%o3) [f(y, x)]
I will check if we have code in core or share, which depends on this
behavior of the function.
Dieter Kaiser