Re: True or False ?

[Austin Obyrne <austin.obyrne@hotmail.com>]

On Sunday, June 17, 2012 3:14:48 PM UTC+1, Martin Trenkmann wrote:
> On 06/17/2012 12:35 PM, adacrypt wrote:
> > On Sunday, June 17, 2012 10:37:17 AM UTC+1, Ludovic Brenta wrote:
> >> Austin writes on comp.lang.ada:
> >>> Do you know if these sort programs are used in real world programming
> >>> today ?
> >>
> >> In our two-million-line mission-critical software we use heap sort and
> >> tree sort but not quick sort.
> >>
> >> --
> >> Ludovic Brenta.
> >
> > Many thanks.
> >
> > I have never heard of heap save - is this to be found in academic books?
> > I'm taking it then that there has not been any huge advance in sorting methods over the past twenty years.
> >
> > Could I pick your brains a bit further.
> >
> > My scheme is ideal for accessing vast programs of millions of lines of source code like you mentioned - I tag every variable as it is being keyed in at the outset (at creation time) and then disable it until it is needed (I comment it out ) and if I do need it I simply uncomment it for sorting by a specially developed sorting method.  The tags can be stored up front or even stored in a file in memory for systematic calling by the main program
> >
> > Perhaps you would have a look at this new method later when I go public.
> >
> > Question - would it call it a big asset to improve on sorting methods? - given that there is so much computer power available to day - poor or even bad methods are getting by without notice?
> >
> > Many thanks for your help again.
> >
> > Regards - Austin.
> >
> 
> Have you done an extensive research of sorting algorithms [1] before 
> inventing your scheme or any comparison regarding best/avg/worst case 
> runtime complexity?
> 
> I think there is still much research in this area going on and depending 
> on the real use case many hybrid algorithms exist. For example the C++ 
> std::sort from GCC uses intosort [2] which is a hybrid of quicksort and 
> heapsort. Python and Java 7, as far as I know, use timsort [3], a hybrid 
> of merge sort and insertion sort.
> 
> So I would be careful stating that "classroom" algorithms are not used 
> in industry.
> 
> Anyway, the description of your scheme is not clear to me, but it 
> reminds me a bit of counting sort [4], just to be sure not to reinvent 
> the wheel.
> 
> Regards -- Martin Trenkmann
> 
> [1] http://en.wikipedia.org/wiki/Sorting_algorithm
> [2] http://en.wikipedia.org/wiki/Introsort
> [3] http://en.wikipedia.org/wiki/Timsort
> [4] http://en.wikipedia.org/wiki/Counting_sort

No, I have not done any research on current sort algorithms.

The replies from you and the other readers have brought me up to date however.
I can see now that there is still some research going on and that is what I really wanted to know - it is not a closed book by any means.

My scheme will make all of these other sort programs redundant possibly but I am not clear why in my own mind - is speed still the essence in sort programs?  I think you will like the connection with Ada when it comes.

I will of course watch out now for making staements about academia.

Could I ask you to watch this space so to speak when I go public with my scheme - there is alot of Ada in it and it may even become a bench mark for comparisons with the existing sort algorithms.

Many thanks for your advice - Austin