On 01/24/2013 12:51 PM, Rainer Weikusat wrote: > linked-list implementation will be good enough because [...] > I don't think it should be discarded as > 'not good enough' based on assumptions alone.
Why not? If a simple model shows significant overhead, and there are no known reasons to doubt that model, why is that not a sufficient justification to try alternatives that have lower overhead? Alex.
