Re: Problem with scaled durations

[David Kastrup]

Oscar Dub <address@hidden> writes:

> Hi everyone,
>
> There appears to be an issue with long lists of differently scaled LilyPond 
> notes where they fail to display, but leave no trace of a problem in the log 
> (example below). I'm having difficulty condensing the issue into something 
> simple enough to send to the bug list, but maybe someone with more insight 
> into how LilyPond calculates time intervals internally can help out.
>
> I thought the problem had something to do with using too many different 
> denominators in the fractions used by the scale syntax. To test this idea 
> out, I made a script generating notes scaled by multipliers from 1/1 to 1/n 
> (output for n=30 shown below).
>
> Running my file, LilyPond only manages to render up to note 18.

You calculate 1/4*(1 + 1/2 + 1/3 + 1/4 ... + 1/18)

If we take a look at the result, it is 14274301:16336320.  Now
multiplying with 1/(4*19) gives 751279:65345280 and the intermediate
calculation has a denominator of 1241560320 before coming to shortest
terms, which is above 2^30.

Guile rationals can deal with arbitrary precision, but LilyPond has its
own idea of rational numbers.

Just try to use numbers that actually add up to somewhat saner values.

-- 
David Kastrup