[Mingw-notify] [ mingw-Patches-2829951 ] Issues with stdint.h header

[SourceForge.net <no...@so...>]

Patches item #2829951, was opened at 2009-07-31 10:11
Message generated for change (Comment added) made by dannysmith
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=302435&aid=2829951&group_id=2435

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: runtime
Group: Patch needs review
Status: Open
Resolution: None
Priority: 5
Private: No
Submitted By: François-Xavier Coudert (coudert)
Assigned to: Chris Sutcliffe (ir0nh34d)
Summary: Issues with stdint.h header

Initial Comment:
GCC-4.5 has good coverage of the C99 types in stdint.h, and it currently has three failures on mingw32 (gcc.dg/c99-stdint-{1,5,6}.c), due to two issues:

  1. int_fast8_t is defined as "char" instead of being "signed char" (as are int_least8_t and int8_t); this is potentially troublesome if one plays with char signedness; also, this is contrary to all headers I could find (except in those that are copied from the mingw32 one!)

  2. the INT8_C, INT16_C, INT32_C, UINT8_C, UINT16_C and UINT32_C macros don't have the right type; they should have simple "int" type, except for UINT32_C which should have "unsigned" type; some explaination of why this is so can be found here by GCC's C99 "resident expert" (http://sourceware.org/ml/newlib/2008/msg00663.html)

With the attached patch, all GCC C99 tests pass.

----------------------------------------------------------------------

>Comment By: Danny Smith (dannysmith)
Date: 2009-07-31 16:03

Message:
I think your changes have already been incorporated into CVS as of
2009-04-11 Danny Smith <dan...@us...>

	* include/stdint.h (int_fast8_t): Specify as signed char.
	(INT8_C, UINT8_C, INT16_C, UINT16_C): Simplify: just allow promotion to
int.
	(INT32_C, UINT32_C): Remove 'L' type specifier on constant.

----------------------------------------------------------------------

Comment By: Chris Sutcliffe (ir0nh34d)
Date: 2009-07-31 14:26

Message:
I'm curious, for the second part of the patch, why have the UINT* macros
lost their unsigned attribute?

For example, looking at the link you specified:

/** Macros for minimum-width integer constant expressions */
 #define INT8_C(x)	x
+#if __STDINT_EXP(INT_MAX) > 0x7f
+#define UINT8_C(x)	x
+#else
 #define UINT8_C(x)	x##U
+#endif

in your patch, there is no check to see what the INT_MAX is, you simply
set it to return val.  Is this correct?


----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=302435&aid=2829951&group_id=2435