Loop unroll fixes

[Bernd Schmidt]

On Fri, 14 Sep 2001, David Edelsohn wrote:

> 	Another GCC developer did respond to my initial request to review
> the patch.  He asked that the patch not be applied until he had reviewed
> it, although he did not have any specific objection when asked.  That
> person has not followed up after initially blocking the patch.

That probably was me, although I recall the details slightly differently.
Of course I can't give a specific objection if I haven't even looked at
the problem.  None of Zoltan's messages contained a good description of
the problem, only testcases - so any potential reviewer would have to
debug the problem and figure out what's going wrong.

> 	I have reviewed the patch.

Then please describe the nature of the bug and how the patch fixes it.

> I have received numerous emails from developers
> extremely frustrated with patches being ignored, not just the complaints
> posted on the GCC mailinglists.

So what do you suggest we do?  Apply patches without understanding them?
I doubt that's going to lead to better quality code.  I think that even
now, patches aren't reviewed extensively enough - a lot of the breakage
we regularly have could be avoided if more time was spent on review.  But
more time is not available.


Bernd