On 11 June 2012 18:38, David Kastrup <d...@gnu.org> wrote: > Well, considering the cost of dynvector-grow!, doing the growth in a > loop rather then just the determination of the new size seems a bit > expensive:
Only if you are repeatedly setting values at indices far beyond the current capacity. This does not occur in my usage patterns where values are primarily inserted at the tail position (i.e. length), which is amortized O(1) since grow! is geometric. This is only an example, something which fits my needs well. You are free to adapt it or not to your particular needs.
