Issue 0EMB.21: [Embedded C 2004 DR#21]

This issue has been automatically converted from the original issue lists and some formatting may not have been preserved.

Authors: WG14
Date: 2005-03-03
Submitted against: Embedded C TR 18037:2004
Status: Fixed
Fixed in: Embedded C TR 18037:2008
Converted from: n1096.pdf, n1180.pdf

Problem: the text on the countls function in 4.1.7.3 and 7.18a.6.4 reads:

The integer return value of the above functions is defined as follows:

• if the value of the fixed-point argument f is non-zero, the return value is the largest integer k for which the expression f<<k does not overflow;
• if the value of the fixed-point argument is zero, an integer value is returned that is at least as large as N-1, where N is the total number of (nonpadding) bits of the fixed-point type of the argument.

Note: if the value of the fixed-point argument is zero, the recommended return value is exactly N-1.

From the definition it is clear that for instance

    countlsur( UFRACT_EPSILON ) == (UFRACT_FBIT - 1)

and

    countlsk ( ACCUM_EPSILON ) == (ACCUM_IBIT + ACCUM_FBIT - 1)

If the text '(nonpadding) bits' is replaced by 'value bits' (see Defect 20), then the text requires that

    countlsr( 0.0r ) >= (N - 1)

where the latter value equals countls( FRACT_EPSILON ). This seems counterintuitive; one would expect the value of countlsr( 0.0r ) to be one less than countls( FRACT_EPSILON ).

Solution: change in 4.1.7.3 and 7.18a.6.4 the text of the 2nd bullet and the Note as follows:

• if the value of the fixed-point argument is zero, an integer value is returned that is at least as large as N, where N is the total number of value bits of the fixed-point type of the argument.

Note: if the value of the fixed-point argument is zero, the recommended return value is exactly N.