/usr/include/curses.h not fixed on sparc-sun-solaris2.5.1

[Gerald Pfeifer]

On Fri, 24 Apr 1998, Per Bothner wrote:
> As I've said before, having sizeof(bool) != 1 is a poor choice.
> It violates users' expectations, wastes space, and as was pointed out in
> a previous discussion, the speed penalty of a one-byte bool can be much
> reduced by some minor Gcc improvements.
> 
> Can we make sizeof(bool) be 1 for -fnew-abi?  Please?

Please keep the performance hit as low as possible!


We have had _quite_ bad experience in a similiar situation:

When switching from the STL from libg++ 2.7.x to the new one from SGI
that ships with egcs, one of our projects had a performance hit of more
than 100%, i.e., our code took more than twice(!) as long due to the new
specialization of vector<bool>. :-(

Gerald
-- 
Gerald Pfeifer (Jerry)      Vienna University of Technology
pfeifer@dbai.tuwien.ac.at   http://www.dbai.tuwien.ac.at/~pfeifer/