incongruency in architecture mismatch linker warnings/errors

[Doug Evans]

Alan Modra writes:
 > > Furthermore, when doing a final link, it succeeds (or gives a warning
 > > when requested), which is counter-intuitive to the user.
 > 
 > True, but one of the original design goals of BFD was to support
 > linking different input formats.

I don't dispute that.  In the context of different input formats
the current behaviour for relocatable links is understandable.
However, the context of my comment above is for the particular
case of same format, same architecture, incompatible machs.
In this particular case it's counterintuitive for the linker to catch
the error during a relocatable link but wimp out with an optional
warning during a final link.

[just want to make sure there's no confusion]