To download the latest version of this file, click on most recent revision number, then use your browser's "save page as" feature.

Use bool instead of int where appropriate.

Print the value of denormals and pseudo-denormals instead of printing NAN as used to be to be compatible with GNU.  This is no longer true.

Revert: "Prototypes of _doprnt, _doscan and _doscan_low moved to libc/doprsc.h"

The rationale behind this changeset was a _doprnt() incompatibility
which had been added to binutils/bfd, but it has already been fixed:
http://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;a=commitdiff;h=8f314ad58ec824ce6c8467af29f11583f79a80ea
Therefore, we don't need this change anymore.

Include libc/doprsc.h.

Update copyright lines of recently modified files

* doprnt.c, doscan.c: locally define bool/false/true and remove the
  stdbool.h usage.

* libc/ansi/time/strftime.c: revert parts of r1.8 and simply use 0
  and 1 instead of true and false again.

* djtar/unbzip2.c: revert parts of r1.4 and go back to the original
  TRUE and FALSE usage.

* tests/libc/c99/math/t-ismac.c: locally define bool/false/true and
  remove stdbool.h usage.

doprnt.c: avoid unused warnings from gcc2

doprnt.c: removed the CHARINT case from the ARG macro.
  (char basetype) seems wrong and the case is handled
  with the generic case already.

Handle case when field width specifier for AEFG conversion specifiers exceeds max default precision of 39 decimal digits.

White space changes.

Revert previous change.

Print sign of NaN.

Minor changes.

Rename exponentl, isspeciall and cvtl into doprnt_exponentl, doprnt_isspeciall and doprnt_cvtl to be consistent with doprnt_roundl.

Fix unused-but-set error/warning.

Use true and false from stdbool.h.

"type punning" fix to pacify gcc 4.3.0.

Implementation of numeric conversion specifier support.
New function __traverse_argument_list added.

Implementation of %[aA] conversion specifier.

Use of nan/NAN and inf/INF (C99 strings) instead of Inf and
NaN strings.  Also return nan/NAN instead of "Unnormal".

New function __grouping_format.  Formats the integer portion
of the result of a decimal conversion (%i, %d, %u, %f, 	%F,
%g and %G) using the non-monetary thousends' grouping character.

Under %+f and the likes, print +nan for NaN,
else print never a sign.

Under %f and the likes, print -0.0 for a negative zero
and -nan for negative NaN.

The configure script of m4-1.4.10 tests certain printf behaviour issueing a
code line similar to this one:
  printf("%010f", 1.0 / 0.0);
Linux produces and passes the test with this output:
  "       inf"
but current CVS version of _doprnt() produces this output:
  "0000000Inf"
The following patch should fix the issue.

Regards,
Juan Manuel Guerrero

Various small fixes to avoid warnings with gcc-4.0 and gcc-4.1. It would
have be nice to commit this in smaller pieces, but it would require
too much work.

Patch from Markus F.X.J. Oberhumer <markus@oberhumer.com>
(http://www.delorie.com/djgpp/mail-archives/browse.cgi?p=djgpp-workers/2006/06/12/06:28:05)

roundl -> doprnt_roundl

Locale fixes from Alexander Aganichev

Change integer conversion for better accuracy.

Build fixes for gcc 3.3 from Andrew Cottrell: type-punning
(Previous commit was done using a slightly old patch. Update
to the newer version, to avoid namespace pollution.)

gcc 3.3 build fixes from Andrew Cottrell: type-punning

Add support for j, z and t conversion specifiers; mark new C99
conversion specifiers as being C99.

Move stdiohk pull in to exit where it is used

Add support for the 'hh' conversion specifier for chars

Update copyright messages

Avoid GCC 3.03 warnings.

Do not mix signed and unsigned.

Under %+f and the likes, print -0.0 for a negative zero.
Reported by Eric Rudd.

Rename nan -> nan_p, since the former conflicts with a function
from libm.a (whose prototype will be added to math.h).

Make sure arguments to ctype functions are always in the
range [-1..255].

update copyright dates

This is a patch for doprnt that prevents it from crashing on unnormals from Eli

import djgpp 2.02

import djgpp 2.01

import djgpp 2.00