Hi Charles.

12-Aug-03 15:33:08, Charles Angelich wrote to Jasen Betts


 CA> 1237c9711cf0
 CA> c_echo



 CA> Hello Jasen - 

 CA> --8<--cut

JB>>>> try designing something that can do,everything the dos
JB>>>> copy command can do (well the things it's documented as
JB>>>> being capable of) without it using any switches...

CA>>> There are so many clones of "copy" I never thought to
CA>>> write another one but I have written a dropin replacement
CA>>> for "sort" in all C code.

JB>> same.... limited to available meory but atleast it uses an
JB>> Order(n.log(n)) sort algorithm (qsort() actually) (which is
JB>> a vast improvemnt on the n-squared algorithm in the dos one.

 CA> Not clear on what "same" you refer to here.

I did the same thiung.

 CA> The definitive DOS sort was
 CA> already written, RPSORT, in ASM and can use virtual memory
 CA> (hard drive) if necessary.

I didn't know that.... thanks.

 CA> I guess that describes it but as I said, I used the K&R qsort
 CA> and I _think_ I wrote my own function to compare the strings?

 CA> It's been some years ago now and I haven't even looked at my
 CA> sort code for many of those years.  I did write a similarity
 CA> sort some years back but that is a horse of a different color.

I remember that thread....  I didn't participate as trying to compreess N
dimensions into 1 seemed like an impossible task to me... I still have your
and Bills source I think...

 CA> We've had many interesting discussions of sorting algorithms
 CA> here in the c_echo not long ago and Bill Buckels did a really
 CA> good job writing his similarity sort routines with a little
 CA> prodding from me. Of the three sim-sorts I know of mine is
 CA> still the fastest but Bill's is the most capable for larger
 CA> files. :-)

my only contribution to sorting algorithms was a way to apply a radix
sort (which is order N and thefore faster than qsort etc on lkarge lists)
to floating point data... dunno if anyone reembers that.

---