On 5/27/26 11:58 AM, Harald Anlauf wrote:
Hi Jerry!
Am 27.05.26 um 3:12 AM schrieb Jerry D:
The attached patch with assistance implements the various rounding modes for
the WRITE with EX hexadecimal float output.
This continues the effort to get the full EX support completed. A followup
patch will take care of the READ EX portion.
This has been regression tested on x86_64.
OK for mainline?
This looks good to me, so good for mainline.
I see one issue when compiling on x86_64-pc-linux-gnu: the snprintf
in libgfortran/io/write_float.def has a mismatch between the format
and the passed parameter (llX == long long unsigned int vs.
GFC_UINTEGER_8). I tried to locally cast the passed value, see
attached patch. This got rid of the warning here.
Committed with the above change using long long unsigned:
The master branch has been updated by Jerry DeLisle <[email protected]>:
https://gcc.gnu.org/g:aacccc01d2bb3dd76ee37281b9f2f04aa9c13f5c
commit r17-848-gaacccc01d2bb3dd76ee37281b9f2f04aa9c13f5c
Author: Jerry DeLisle <[email protected]>
Date: Sat May 23 20:28:43 2026 -0700
Fortran: [PR93727] Add EX format rounding for truncated hex mantissa
