Re: Alternative to linewidth=-1

[Rune Zedeler]

Han-Wen wrote:

> on the whole (well perhaps more O(n log n + k n^2 ) with a very small

Oh yes, but O(n log n) is also slow compared to O(n). (I remember
replacing a <set> with a <vector> on a simple one million elements sieve
of erathostenes. I got a facter 20 speed up - 2 seconds instead of 40 -
because log_2(1000000) = 20)

Are there other parts of lily than the spacing that is nlogn (if we
asssume the same amount of variables,etc)?

-Rune