Hi John,
You wrote to me:
JG> GW> The sort you use is a customised version of one Jonathan de Boyne
JG> GW> Pollard calls a "comb sort", he posted sample code over in C_Plusplus
JG> GW> recently. It stacks up well in general use, in my testing a comb sort
JG>It is in fact a slightly modified Shell-Metzner. Giving it a different 
ame
JG>changes nothing. Any minor performance variation which can
JG>be seen compared to
JG>a "pure" Shell-Metzner is due to it being tailored to the particular task.
OK. It's not one I'd come across in that form, so I used the name the
person who posted it referred to it by.
JG>There is a fair bit of this name changing going on around here at the 
moment
JG>It would appear some people don't even recognise a simple
JG>Bubble sort when they
JG>see it and insist on calling it by a different name simply because when
JG>properly coded it far exceeds their expectations. *I* know what I'm 
oding,
JG>even if others don't. The fact that a particular sort may demolish the
JG>"mathematical proof" is because of a basic flaw in the mathematics, not 
he
JG>sort. For such reasons I'll now bow out of this discussion.
I _do_ recognise a simple Bubble sort. If coded in the way you consider
"proper" I do not consider it a Bubble sort as it does not compare
ajacent elements, which is the thing that defines a bubble sort as far
as I'm concerned. I agree that changeing the sort to the way you like
gives far better performance, but makes it an example of a different
type of sort.
George
 * SLMR 2.1a * Wastebasket: Something to throw things near.
--- Maximus/2 3.01
---------------