[patch, libgfortran] PR32446 F0.n output format fails with large numbers

[Jerry DeLisle]

:ADDPATCH fortran:

This patch fixes this by setting the ndigits to zero appropriately.  If you 
study the code you will see that previously, ndigits was going negative for the 
case where the nbefore is greater.  The initial patch I posted to the PR was way 
wrong. :)

Regression tested on x86-64-pc-Gnu/Linux.

Test case included.

OK for trunk?

Eventually, this should go to 4.2.  It is a regression against g77.  I was able 
to reproduce the failure with large width formats and large real constants and 
they passed on g77.

Regards,

Jerry

2007-06-22  Jerry DeLisle  <jvdelisle@gcc.gnu.org>

	PR libgfortran/32446
	* io/write.c (output_float): Calculate ndigits correctly for large
	numbered formats that must pad zeros before the decimal point.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: fmt_huge.f90
Type: text/x-fortran
Size: 179 bytes
Desc: not available
URL: <http://gcc.gnu.org/pipermail/fortran/attachments/20070623/84b014d7/attachment.bin>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: pr32446.diff
Type: text/x-patch
Size: 901 bytes
Desc: not available
URL: <http://gcc.gnu.org/pipermail/fortran/attachments/20070623/84b014d7/attachment-0001.bin>