libstdc++/4150: catastrophic performance decrease in C++ code

[Loren James Rittle]

In article <wvlelhbgbwb.fsf@prospero.cambridge.redhat.com>,
Jason Merrill <jason@redhat.com> writes:

> Why would overflow ever need to call fflush?

I have no good answer to that question.  Thus, I have attempted to
remove it ever since it was stuck in there.  However, every attempt I
have made has broken one or more test cases (in dejagnu or the
interactively run cases such as the scott snyder test).

Regards,
Loren