lib/w64devkit
1
0
Fork 0
mirror of https://github.com/skeeto/w64devkit.git synced 2026-10-23 06:08:56 +03:00
Code Activity

Import Peter0x44's printf %g patch

Browse Source
This commit is contained in:
Christopher Wellons
2025-10-11 09:48:39 -04:00
parent 53cab45eb1
commit 308f5abc19
2 changed files with 55 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
Dockerfile
View File

@@ -137,7 +137,9 @@ RUN mkdir -p $PREFIX/lib \
&& ln $PREFIX/lib/libchkstk.a /bootstrap/lib/
WORKDIR /x-mingw-crt
RUN /mingw-w64-v$MINGW_VERSION/mingw-w64-crt/configure \
COPY src/crt-printf-g.patch $PREFIX/src/
RUN patch -d/mingw-w64-v$MINGW_VERSION -p1 <$PREFIX/src/crt-printf-g.patch \
&& /mingw-w64-v$MINGW_VERSION/mingw-w64-crt/configure \
--prefix=/bootstrap \
--with-sysroot=/bootstrap \
--host=$ARCH \

52
src/crt-printf-g.patch Normal file
View File

@@ -0,0 +1,52 @@
commit 8612006d7f5376cdefa7c887db6d3dfb48d285be
Author: Peter Damianov <peter0x44@disroot.org>
Date: Wed Oct 8 11:12:52 2025 +0100
Fix %g format specifier width padding with trailing zeros
Remove incorrect width adjustment that prevented proper padding
with spaces for %g format specifier when the value has trailing
zeros in the integer part.
The original code would reduce the field width when precision
became negative (e.g., for 100000.0, precision = strlen("1") - 6 = -5),
causing the output to lose its padding:
printf("(%10g)\n", 100000.0f); // was: (100000), expected: ( 100000)
The fix clamps negative precision to 0 instead of adjusting the
width, preserving the requested field width for proper padding.
Adds test case t_printf_g_width.
Reference: https://github.com/skeeto/w64devkit/issues/269
Signed-off-by: Peter Damianov <peter0x44@disroot.org>
Signed-off-by: LIU Hao <lh_mouse@126.com>
diff --git a/mingw-w64-crt/stdio/mingw_pformat.c b/mingw-w64-crt/stdio/mingw_pformat.c
index b7d3bfb41..3df14704b 100644
--- a/mingw-w64-crt/stdio/mingw_pformat.c
+++ b/mingw-w64-crt/stdio/mingw_pformat.c
@@ -1881,14 +1881,14 @@ void __pformat_gfloat( long double x, __pformat_t *stream )
* precede the radix point, but we truncate any balance following
* it, to suppress output of non-significant trailing zeros...
*/
- if( ((stream->precision = strlen( value ) - intlen) < 0)
- /*
- * This may require a compensating adjustment to the field
- * width, to accommodate significant trailing zeros, which
- * precede the radix point...
- */
- && (stream->width > 0) )
- stream->width += stream->precision;
+ stream->precision = strlen( value ) - intlen;
+ /* When the mantissa is shorter than the number of integer digits
+ * (e.g., 100000 has mantissa "1" but requires 6 digit positions),
+ * precision becomes negative. Clamp to zero to represent no
+ * fractional digits.
+ */
+ if( stream->precision < 0 )
+ stream->precision = 0;
/* Now, we format the result as any other fixed point value.
*/