Readme file for CLIB V3.34O

Contents

·         Important Information

·         Manual Corrections and Updates

·         Known Problems in Current Version

·         Program Corrections and Updates

Important Information

This readme file contains information that are valid for two products, the normal package included with a compiler (all standard include files and some c files) and the whole source for the library.

3.34D

o        In previous versions of this file, some bugs in section 4 were put under the version for which they were reported. Now, all bugs (for all listed versions) are described under the version in which they were fixed.

3.34A

o        printf, CHANGED BEHAVIOR:

When using REDUCED_SUPPORT (_small_write) in _formatted_write, %X, %x and %o previously zero-padded the output. Now, leading zeros are removed.

3.33F

o        printf, CHANGED BEHAVIOR:

When using REDUCED_SUPPORT in _formatted_write, %x previously printed capitals, e.g., printf("%x", 0x1f2a); outputs "0x1F2A". While correcting this, support was included for %X to allow the same output as before. h and l modifiers are also supported in REDUCED_SUPPORT but can easily be removed when not necessary.

Manual Corrections and Updates

3.33C

o        The exp10 function is not described in manuals produced before 8th of November 1996.

o                            Include file:    math.h
o                            Declaration:     double exp10(double arg)
o                            Parameters:      arg    A double value
o                            Return value:    a double with the value 10^arg
Description:     Computes the value 10^arg

Known Problems in Current Version

Program Corrections and Updates

3.34O - 2003-11-14

o        ptrdiff_t was changed into a signed integer type (int or long).

3.34N - 2003-03-06

o        Added stdbool.h.

3.34M - 2002-12-06

o        frmrd.c - Handles %*n correctly now [EW11577].

3.34L - 2001-06-15

o        sysmac.h - __IAR_SYSTEMS_ICC__ set to 5 now allowed.

o        stdlib.h, calloc.c - First parameter changed from unsigned int to size_t.

3.34K - 2000-10-06

o        heap.c
The rebuild info has been deleted.

3.34J - 1999-09-16

o        setjmp.h
Both 'setjmp' and 'longjmp' must be declared as __standard_function if the target is H8/300.

3.34I - 1999-08-16

o        strtok.c
Fixed it when fed with a string like "12|34\0567". If searching for "|" it then returned "12", "34", and "567".

o        frmwri.c (REDUCED variant)
Added a cast in the va_arg read.

o        assert.h
Deleted the warnings pragma settings.

3.34H - 1999-05-20

o        stdarg.h, va_start() for ICC7700, ICC8051 ver 1 - 3.10 - Inserted a cast to void *

o        frmwri.c (small_write version) - %d for negative values now prints correctly.

o        strtol.c, strtoul.c - Overflow is now correctly identified.

3.34G - 1999-02-15

o        atol.c, atoi.c, atof.c, free.c - Slight optimization.

o        ctype.h, iccext.h, icclbutl.h math.h, setjmp.h, stdio.h, stdlib.h, string.h - Deleted parameter names on function declarations.

o        All files - Style changed whereever applicable, tabs removed.

3.34F - 1998-11-24

o        malloc.c,realloc.c,heap.c,memstruc.i - Deleted align_on macro, and rewrote alignment code.

o        malloc.c, realloc.c - Changed style of coding.

3.34E - 1998-11-16

o        sysmac.h - __IAR_SYSTEMS_ICC__ set to 3 now allowed.

o        wchar_t in stddef.h is now set to __WCHAR_T_TYPE__, if defined.

3.34D - 1998-09-18

o        CLIB0017: scanf: frmrd.c - frmrd ignored formatter string L-modifier when int was larger than short.

o        CLIB0016: malloc.c - PTR_ATTRIBUTE was missing in align_on macro.

o        CLIB0015: heap.c - PTR_ATTRIBUTE replaced by MEM_ATTRIBUTE for "char bulk_storage[MALLOC_BUFSIZE]".

o        exp.c, log.c - test __IAR_SYSTEMS_ICC__ instead of __IAR_SYSTEMS_ICC.

o        qsort.c - QSORT_MAXSIZE can be defined externally.

o        scanf.c - SCANF_MAXLINE can be defined externally.

o        ctype.h, iccext.h, iccbutl.h, stdarg.h - Includes sysmac.h.

o        ctype.h, iccext.h, iccfloat.h, icclbutl.h, math.h, setjmp.h, stdarg.h, stdlib.h, string.h, sysmac.h - Uses __INTRINSIC macro.

o        stdarg.h - Test for __IAR_SYSTEMS_ICC__ >= 2 (before == 2).

o        stdio.h - Removed unused pragma.

o        stdio.h - Removed sysmac.h include.

o        sysmac.h - Corrected definition of __IAR_SYSTEMS_ICC__.

3.34C - 1998-07-01

o        CLIB0014: math.h, tan.c - Confusing constant name: '_INVPI' = 4/pi, changed to 1/pi.

o        CLIB0013: strtol.c,strtoul.c - When string contained zeros only, e.g., "0000", **endptr was '0', not '\0' as it should.

3.34B

o        stdarg.h - __va_start() supplied by some compilers. Two flavors of va_arg() available.

o        stddef.h - Changed definition of offsetof() macro.

3.34A - 1998-04-31

o        CLIB0012: printf: frmwri.c (_small_write), intwri.c - The small formatter did not output the correct value for 32 bit numbers when using %o.

o        CLIB0011: qsort.c - The internal buffer pointer pointed wrong when calling qsort more than once.

o        CLIB0010: printf: frmwri.c (_small_write) - The small formatter zero-padded output when using %x, %X and %o.

o        sysmac.h is a new include file that governs target-dependent options in the library. A lot of files need to include this.

o        isxxx.c - #pragma function=intrinsic(0) removed.

o        cosh.c, exp.c, exp10.c, log.c, pow.c, satan.c, sinh.c, sinus.c, sqrt.c, tan.c, tanh.c, float.h, iccfloat.h, math.h - Uses __FLOAT_SIZE__ and __DOUBLE_SIZE__ now. float.h uses __LONG_DOUBLE_SIZE__ as well.

o        limits.h - Uses __INT_SIZE__. Max and min supplied by compiler (e.g. CHAR_MAX...).

o        setjmp.h - Uses __JMP_BUF_ELEMENT_TYPE__ and __JMP_BUF_NUM_ELEMENTS__.

o        stddef.h,stdio.h,stdlib.h,string.h - size_t and ptrdiff_t defined in sysmac.h.

o        frmrd.c,frmwri.c,intwri.c - Uses VAPTR macro now.

3.33H - 1998-01-19

o        CLIB0009: scanf: frmrd.c - sscanf("ABC", "%s%n", myStr, &myInt) returned with myInt == 0, i.e., zero characters scanned when reaching the %n.

o        printf: frmwri.c (reduced_support) - Modified va_arg() fetch for h-modifier to remove 32-bit target bug.

3.33G - 1997-09-10

o        printf: frmwri.c - Detail improvements to ANSI compliance:

+ ("%0*d",-3,0) produced "000" instead of "0 ".
+ ("%#.0o",0) should produce one zero.
+ unsigned conversions did not ignore the "+" flag.
+ unsigned conversions did not ignore the " " flag.
+ ("%#.2o",1) padded one extra zero.

o        CLIB0006: printf: frmwri.c - Zero field width supplied as a argument, printf("%*.0f", 0, 1.654), will not produce any output. The correct output should be "1".

o        CLIB0007: strtod.c - When string contains zeros only, e.g., "0000", **endptr will be '0', not '\0' as it should.

o        CLIB0008: stdarg.h - Some of the older IAR compilers for chip with word alignment on stack generated a faulty address in some cases if an 8-bit object was given as the parameter to the 'va_start' macro.

3.33F - 1997-05-27

o        CLIB0005: printf: frmwri.c - Zero precision, printf("%.0d",0), prints out a "0" instead of nothing. ANSI says that if the argument is zero then nothing should be printed if the precision is zero.

o        printf: frmwri.c (reduced_support) - %x and %X added. h and l modifiers added.

3.33E - 1997-04-17

o        Some minor adjustments in 'stdarg.h'.

o        An uninitialized variable in 'libtest.c' could make the test crash.

3.33D - 1997-03-20

o        stdio.h - Two new functions have been added 'vprintf' and 'vsprintf'.

o                            vprintf : Works as 'printf' with the variable argument list of
o                                      'printf(...)' replaced with a 'va_list' argument.
o                             
o                            vsprintf: Works as 'sprintf' with the variable argument list of
          'sprintf(...)' replaced with a 'va_list' argument.

For more information about these functions, see the ANSI standard definition.

o        CLIB0002: stdarg.h - The function/macro 'va_end' does now return ((void)0) as default.

o        CLIB0003: printf/scanf: frmwri.c, frmrd.c - When using pointer formatter ( %p ) the default format for scanf was decimal number and the default format for printf was hexadecimal number. The scanf has been changed to have hexadecimal number as default.

o        CLIB0004: scanf: frmrd.c - 'scanf' assumed that 'short' was the same as 'int' and therefore scanned a 32-bit value when the formatter "%h" was given.

3.33C - 1996-12-19

o        printf: (and releated functions) - The printf functions default precision is set to 6 digits. In earlier version printf could write seven or more digits.

o        printf:(and releated functions) - When:

- the conversion specifier is d or i
- the field with is set to zero
- the value to be printed is zero

then there should be no character written. In earlier version a zero will be written but not in this version.

o        scanf:(and releated functions) - When scanning for a float/double number the buffer pointer would not be updated to point out the next character to be read. The pointer still point out the last character read. This means that the last digit in the float/double number will the next character to be read.

o        math.h: - The declaration of exp10 is missing. Should be: MEMORY_ATTRIBUTE double exp10 (double __arg);

 

$Revision: 1.9 $

Copyright 2000 IAR Systems. All rights reserved.