[r6rs-discuss] [ANN] scheme-reports.org

[Brian Mastenbrook]

On Aug 25, 2009, at 5:08 PM, David Rush wrote:

> Well actually, after spending some time reading the Stalin source code
> (and seeing how well it optimizes this kind of thing), I actually went
> back to using lists of characters for a large number of string
> algorithms. It really is a very graceful way to handle strings -
> substring, insert, and delete all become *very* cheap. It's a paradigm
> I'm quite sure that we ought to preserve.


I'm not familiar with the Stalin optimizations you describe, but why  
is a list of characters preferable to a circular gap buffer? The worst  
case performance of a gap buffer is much better than that of a linked  
list of characters.
--
Brian Mastenbrook
brian at mastenbrook.net
http://brian.mastenbrook.net/