[patch, libgfortran] PR32611 Print sign of negative zero

[Janne Blomqvist]

On 7/15/07, Jerry DeLisle <jvdelisle@verizon.net> wrote:
> FX Coudert wrote:
> > Hi Jerry, hi Janne, hi all!
> >
> >> I have provided a compile time flag, -fsign-zero that enables the
> >> negative sign on zero values.  This is the default behavior.
> >> -fno-sign-zero disables the negative sign.  Though probably never
> >> used, it is there for anyone who comes upon an old application where
> >> it might matter.
> >> I plan to commit the patch today or tomorrow.  The patch will include
> >> fixes for test cases already submitted in the first patch I gave in
> >> this thread.
> >
> > I've sent my previous answer to your patch too fast, and after that I've
> > thought some more. I've read your patch, and it's OK to me if you add
> > documentation for the new option. Still, I think it would really be
> > great to integrate it with my patch changing the ABI for passing
> > compile-time options. To be completely honest, I am even more interested
> > in merging these patches that I will have intermitent (unreliable)
> > internet access during the next few days, and then will be offline for a
> > week in the beautiful but desert coutryside of Ardèche (returning July
> > 27th).
> >
> > Thus, I've merged the two patches together and written a basic ChangeLog
> > (feel free to edit it). If you are OK with my part of the patch, and let
> > 48 hours for others (including Janne) to comment, then I think we (and
> > by this "we", I really mean "you" ;-) can safely commit the patch,
> > provided you write the little piece of missing documentation.
> >
> > Of course, if you feel that the task I'm imposing upon you is too much
> > of a burden, please feel free to commit your patch by itself and let me
> > worry later about the ABI change.
> >
> > Cheers,
> > FX
> >
> >
> >
> OK, I will do the documentation part and commit.  Will wait for Janne and others
> to comment.  How about 24 hours, since by the time 48 happens, I am at work
> again.  :)

Sorry I didn't get around to doing a "proper" review of FX's patch
like I promised. I read through it, and the new patch, and  they look
fine, although I haven't applied it and regtested.  I few minor style
issues in libgfortran/io/write.c:

- When calling calculate_sign, space between the function name and "(".

- Update the comment to reflect that the option is -fsign-zero and not
-std=legacy.

-- 
Janne Blomqvist