1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
|
/* ----------------------------------------------------------------------------------------------
strtod.s
Copyright (c) Michael Stumpf <mistumpf@de.pepperl-fuchs.com>
addition to Floating point library for Atmel AVR uC
converts a string to a double ( = float)
double := [ws]*[+|-][0123456789]*[.0123456789]*[e|E][+|-][0123456789]*
Author : Michael Stumpf
mistumpf@de.pepperl-fuchs.com
Version : 0.1.0
Date : 17.09.1999
Algorithm:
-0123456789.01234567e+012
converted to
unsigned long integer fraction :
ul = 1234567890 (=0x49 96 02 D2) with
(max 4 byte, breaks if uppermost nibble != 0)
fraction power (rFracPower) = 9 and
decimal point shift (rDPshift) = -1
integer exponent : rExp = 12 rDPshift-1
then if no overflow/underflow occurs the fp number is
(corrected by decimal point shift)
fp = (float)ul * 10^(rExp+rDPshift)
overflow occurs if exp>0 && exp+fraction power > 38
or
fp results in NaN
fp = +/-MAX_FLT then, errno = ERANGE (if __ERRNO__ is defined)
underflow occurs if exp<0 && |exp|-fraction power > 38
or
fp results in zero and exp<0 && |exp|-fraction power > 38
fp = 0.0 then, errno = ERANGE (if __ERRNO__ is defined)
Exceptions: no NAN INF IND strings are recognized
================================================================================
Version Date Changes
0.1.0 * first release
0.1.1 02.11.99 * macro mask(bit) replaced by BV(bit)
|
