a # 1 65 R ' z # n alprec x                                                  '
a # a,[1]  '"alprec - alter precision of a scalar or multiprecision number.  '
a # a,[1]  '"from ''the apl handbook of techniques'' copyright 1978, ibm.      '
a # a,[1]  '"-for x a scalar or multiprecision integer or floating point     '
a # a,[1]  '" number of precision m (i.e. (rx)=m+1), z will be set to the    '
a # a,[1]  '" floating point number of precision m+n whose value is the      '
a # a,[1]  '" same as or truncated from that of x.  (scalars have            '
a # a,[1]  '" implicit precision 3.)                                         '
a # a,[1]  '"-in brief, x is extended on the right with zeros (if n>0)       '
a # a,[1]  '" or truncated on the right (if n<0).  the exponent (1yx) is     '
a # a,[1]  '" increased by n to compensate.                                  '
a # a,[1]  '"-since all the functions in the multiprecision floating         '
a # a,[1]  '" point package determine the precision of the result from       '
a # a,[1]  '" the precisions of the input, it is often necessary to          '
a # a,[1]  '" indicate the precision of some exact quantity (such as an      '
a # a,[1]  '" integer) so as not to cause involuntary shortening of          '
a # a,[1]  '" precision.  see the examples below.                            '
a # a,[1]  '"-decreasing the precision of a number (by using negative n)     '
a # a,[1]  '" may occasionally be necessary if the result of an inter-       '
a # a,[1]  '" mediate calculation is assigned a spuriously high              '
a # a,[1]  '" precision by one of the arithmetic functions.                  '
a # a,[1]  ' z#(_n_1Yx),1U((R,x)+n#nS2_R,x)Yx#float x                        '
Lfx a