Re: Support of log10?

[Peng Yu]

> | 16.3 Arbitrary-Precision Arithmetic Features in gawk
>
> | By default, gawk uses the double-precision
> | floating-point values [...]  However, if it was
> | compiled to do so, and the -M command-line
> | option is supplied, gawk uses the GNU MPFR and
> | GNU MP (GMP) libraries

The above sentence is not clear. What is the difference between MPFR
and GMP? Why both of them need to be used?

> for arbitrary-precision
> | arithmetic on numbers.
>
> > echo 3 | gawk -M -v 'OFS=\t' -v OFMT=%.20g           \
>      -e 'function log10(x) { return log(x)/log(10); }  \
>      { print "log10(" 10^$1 ")", log10(10^$1) }'
>
> log10(1000)     3

The drawback of -M is that it can be many times slow? In this sense
float log10 function may still be useful?

time gawk -e 'function log10(x) { return log(x)/log(10); } {
for(i=1;i<=1000000;++i) log10(10^$1) }' <<< 3

real    0m0.180s
user    0m0.176s
sys    0m0.002s
time gawk -M -e 'function log10(x) { return log(x)/log(10); } {
for(i=1;i<=1000000;++i) log10(10^$1) }' <<< 3

real    0m2.842s
user    0m2.838s
sys    0m0.003s

> See also footnote 98 and its referencing text.

Where is footnote 98?

-- 
Regards,
Peng