[LAU] OT(ish): Strange coding problem (audio related)

[James Stone]

On Fri, Jan 28, 2011 at 12:03 PM, Peter Nelson <peter at fuzzle.org> wrote:
> On Fri, 2011-01-28 at 12:28 +0100, Philipp Überbacher wrote:
>
>> Interesting.. would you mind explaining how this can be?
>> How can 11-11 yield 4.80518e-16?
>
> Because, rather than 11, log(2048)/log(2.0) is actually (to 40 d.p.)
>
> 7.6246189861593984035895533360399422488305 /
> 0.6931471805599453094172321214581765680755
>
> But doubles don't contain 40 d.p.; it is approximately 16. Thus some
> rounding errors may occur...
>

.. I wonder if this bugfix should be added to some other projects? I
think the same fft.h is used in one of the functions in audacity?
Perhaps someone who is on LAD could cross-post this?

James