Patch to rework IEEE floating point checks

[Joseph S. Myers]

On 28 Feb 2002, Richard Sandiford wrote:

> 3) Document what exactly makes each transformation right or wrong,
>    which should make things easier to maintain.  At the moment, most
>    checks just say "not valid for IEEE" with no explanation why.
>    And to show that there has been some confusion here, if x and y
>    are IEEE floats, the tree level optimises "x + -y" into "x - y",
>    but the RTL level doesn't.
> 
> 4) Make it possible to add options like -fignore-signed-zeros,
>    or whatever, as suboptions of -funsafe-math-optimizations.
>    (More options might not be universally popular, though. ;-)

Another thing that would be desirable: allow tree and RTL expressions to
carry with them flags detailing which transformations are allowed.  (This
is needed for the C99 standard pragmas, and I think it is useful for Java
as well.)

-- 
Joseph S. Myers
jsm28@cam.ac.uk