This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: -funsafe-loop-optimizations
On Jan 2, 2005, at 4:46 AM, Kai Henningsen wrote:
gdr@integrable-solutions.net (Gabriel Dos Reis) wrote on 02.01.05 in
<m3652g8vtz.fsf@uniton.integrable-solutions.net>:
Paul Schlie <schlie@comcast.net> writes:
| therefore if there is
concern
| that a potential circumstance may either likely be unintentional,
and/or
| prevent certain optimizations, then simply warn, nothing else;
where then
| the programmer has the option to either modify the code or not
depending
on | their true desire and understanding of it's implications.
That makes the most sense to me.
From a purely programmer-centric view, what I'd like to be able to do
(and
this is not just for loops) is the following:
1. Assert somehow which calculations are/aren't supposed to
overflow/wrap
(where I care)
One important special case: overflow is undefined behavior for signed
integer types. We have every right to assume that loop counters of type
'int' won't overflow.
--Matt