Fix stupid error in my last change to highest_pow2_factor

[Geert Bosch]

On Monday, December 17, 2001, at 02:00 , Mark Mitchell wrote:
> For the FSF releases, we are obligated to avoid regressions in the
> test suites that ship with GCC.  If new tests are not contributed,
> then we may accidentally break things that were previously fixed.
>
> If it is not worth it to the Ada contributors to create new tests
> that can be checked in to GCC to prevent that from happenning,
> that is OK with me.

I think we should proceed as follows: Laurent is working on integrating
the ACATS tests. These are a good basic set of tests covering most of
the compiler. An Ada compiler that passes all tests without problems,
is in good shape.

Next, we sugs arhould demand full compilable test cases for any Ada bugs 
reported
against FSF releases of GNAT. When these be fixed, the cases should be 
filed as
part of the GCC regression suite. This way, we will both be building a
valuable and unique library of test cases and we will be preventing 
regressions
in bugs reported against FSF releases.

Note that, even though ACT cannot contribute their regression tests, 
before
I approve any non-trivial front-end patches, I do run them against these 
tests.
So, even though the DEC^H^H^HCompaq Ada test suite and ACT regression 
test suites
cannot be integrated in FSF sources, they are run and if any regressions 
do
occur from these, they will be fixed.

With these quality assurances, I do believe we are in very good
shape to ensure the continuing health of the Ada front end.

   -Geert