Serious bug

Craig Burley burley@gnu.org
Fri Sep 25 13:25:00 GMT 1998


>"but a few programs rely on the precise definition of IEEE floating
>point.  Use `-ffloat-store' for such programs."
>
>So -ffloat-store does not work as documented, since some constructs
>still manage to disobey IEEE behavior even with it on.

Actually, it does work as documented, if you take into account the
entire writeup on the option, instead of just looking at the last
few words.  It starts off saying "Do not store floating point variables..."
(though perhaps that should be emphasized as "floating-point
@emph{variables}...").

Nevertheless, good documentation should, IMO, not easily suffer this
sort of misreading, so I propose the following patch.

Also, perhaps it's worth considering providing -ffloat-store-all,
to store intermediate results as well, which will *really* slow
down some programs.  :)

        tq vm, (burley)


Fri Sep 25 12:51:28 1998  Craig Burley  <burley@melange.gnu.org>

	* invoke.texi (-ffloat-store): Clarify that this option
	does not affect intermediate results -- only variables.

*** g77-e/gcc/invoke.texi.~1~   Mon Jul 13 08:39:33 1998
--- g77-e/gcc/invoke.texi       Fri Sep 25 12:50:25 1998
*************** precision than a @code{double} is suppos
*** 2190,2194 ****
  x86 architecture.  For most programs, the excess precision does only
  good, but a few programs rely on the precise definition of IEEE floating
! point.  Use @samp{-ffloat-store} for such programs.

  @item -fno-default-inline
--- 2190,2195 ----
  x86 architecture.  For most programs, the excess precision does only
  good, but a few programs rely on the precise definition of IEEE floating
! point.  Use @samp{-ffloat-store} for such programs, after modifying
! them to store all pertinent intermediate computations into variables.

  @item -fno-default-inline



More information about the Gcc-patches mailing list