OSDN Git Service

2011-05-01 Jerry DeLisle <jvdelisle@gcc.gnu.org>
[pf3gnuchains/gcc-fork.git] / libgfortran / ChangeLog
index 012ebe8..5b73722 100644 (file)
@@ -1,3 +1,76 @@
+2011-05-01  Jerry DeLisle  <jvdelisle@gcc.gnu.org>
+
+       PR libgfortran/48787
+       * io/write_float.def (output_float): Gather up integer declarations and
+       add new 'p' for scale factor. Use 'p' in place of the 'dtp' reference
+       everywhere. For ROUND_UP scan the digit string and only perform
+       rounding if something other than '0' is found.
+
+2011-04-29  Janne Blomqvist  <jb@gcc.gnu.org>
+
+       * io/unix.c (min): New macro.
+       (unpack_filename): Return errno number for errors.
+       (regular_file): Use appropriately sized buffer for path.
+       (compare_file_filename): Likewise.
+       (find_file): Likewise.
+       (delete_file): Likewise.
+       (file_exists): Likewise.
+       (file_size): Likewise.
+       (inquire_sequential): Likewise.
+       (inquire_direct): Likewise.
+       (inquire_formatted): Likewise.
+       (inquire_access): Likewise.
+
+2011-04-29  Janne Blomqvist  <jb@gcc.gnu.org>
+
+       PR libfortran/48488
+       * io/write.c (write_real, write_real_g0): Update comments.
+
+2011-04-29  Jerry DeLisle  <jvdelisle@gcc.gnu.org>
+           Janne Blomqvist  <jb@gcc.gnu.org>
+
+       PR libgfortran/48488
+       PR libgfortran/48602
+       PR libgfortran/48615
+       PR libgfortran/48684
+       PR libgfortran/48787
+       * io/write.c (write_d, write_e, write_f, write_en,
+       write_es): Add precision compemsation parameter to call.
+       (set_fnode_default): Adjust default widths to assure
+       round trip on write and read. (write_real): Adjust call to write_float.
+       (write_real_g0): Calculate compensation for extra precision and adjust
+       call to write_float.
+       * io/write_float.def (output_float_FMT_G_): Use volatile rather than
+       asm volatile to avoid optimization issue. Correctly calculate the
+       number of blanks (nb) to be appended and simplify calculation logic.
+       (write_float): Increase MIN_FIELD_WIDTH by one to accomodate the new
+       default widths. Eliminate the code that attempted to reduce the
+       the precision used in later sprintf functions.  Add call parameter to
+       compensate for extra precision.
+
+2011-04-20  Jim Meyering  <meyering@redhat.com>
+
+       * intrinsics/move_alloc.c (move_alloc): Remove useless
+       if-before-free test.
+       * io/fbuf.c (fbuf_destroy): Likewise.
+       * io/format.c (save_parsed_format): Likewise.
+       * io/open.c (already_open): Likewise.
+       * io/unit.c (free_internal_unit, close_unit_1): Likewise.
+       * io/unix.c (mem_close): Likewise.
+
+2011-04-19  Jerry DeLisle  <jvdelisle@gcc.gnu.org>
+
+       PR libgfortran/48602
+       * io/write_float.def (output_float_FMT_G): Fix reversal in conditional.
+       Use asm volatile to mark temp variable, avoiding optimization errors.
+
+2011-04-17  Jerry DeLisle  <jvdelisle@gcc.gnu.org>
+
+       PR libgfortran/48602
+       * io/write_float.def (output_float_FMT_G): Use current rounding mode
+       to set the rounding parameters. (output_float): Skip rounding
+       if value is zero.
+       
 2011-04-16  Janne Blomqvist  <jb@gcc.gnu.org>
 
        * intrinsics/date_and_time.c (date_and_time): Remove sprintf CPP