On 6 January 2016 at 15:21, Dean Rasheed <dean.a.rasheed@gmail.com> wrote:
> Actually I found the implementation a little confusing, partly because
> the first comment doesn't really match the line of code that follows
> it, and partly because of the complexity of the loop, the macros and
> working out all the exit conditions from the loop. A much simpler
> implementation would be to first call strip_var() ...
Additionally, the central part of the algorithm where it computes dig
% 10, dig % 100 and dig % 1000 inside a bunch of conditional macros is
overly complex and hard to read. Once you've got it down to the case
where you know dig is non-zero, you can just do
while ((dig % 10) == 0)
{ dscale--; dig /= 10;
}
which works for any NBASE, rather than having to introduce an
NBASE-dependent block of code with all those preprocessor
conditionals.
(Actually, given the other thread of this discussion, I would name
that local variable min_scale now.)
Regards,
Dean