Mail Archives: djgpp/1997/05/21/10:49:12
I added comp.oos.djgpp group to this thread, and removed comp.os.linux.advocacy:
NetRunner <s3491939 AT techst02 DOT technion DOT ac DOT il> wrote in article <Pine DOT GSO DOT 3 DOT 95 DOT 970520121658 DOT 11576A-100000 AT techst02 DOT technion DOT ac DOT il>...
> On 21 Sep 1996, Ed Parry wrote:
>
> > Date: 21 Sep 1996 13:22:02 -0700
> > From: Ed Parry <edparry AT primenet DOT com>
> > Newsgroups: comp.lang.c++, comp.lang.c, comp.os.linux.advocacy,
> > alt.lang.basic, alt.destroy.microsoft, alt.fan.bill-gates
> > Subject: Re: Proposal: <basic.h> File
> >
> > nedkelly AT eagle DOT ais DOT net (Ned Kelly) wrote:
[snip]
> Well, both of you sounds quite desperate. That's bad.
> I order to program in C, the whole idea is not to simulate C to Basic, but
> simply to learn C.
^^^^^^^^^^^^^^^^^^ I agree absolutely.
> C is not hard. If you can program in Basic then you can program in C.
Also, that is true, but I must comment this reply:
C is quite different than BASIC and it is not easy to switch from BASIC to C. It seems like switching from market mannagement to field service engineering.
The point is that under C programmer uses quite different programming techniques than in BASIC (e.g.: pointers).
Also, programming in C is not easy -- it is not easy to remove all bugs from program. Under BASIC it __is_ very easy to write bug free application.
Most programmers claim against BASIC because of speed, but this is totaly *wrong*. For example of speed, check speed of my program BPSCAD (http://www.iridis.com/bps/software.htm, http://www.iridis.com/bps/bpscad.htm) how fast it finds particular point in data base by coords: less than second from database of 150000 points on PC/AT286-16MHz without FPU.
Combination of BASIC and assembler gives to me very good results. Also, BASIC seems for me the best tool for development of new algorythms and optimisation methods, e.g.: I made TSR driver for four steppers (microstep controll) in way that I made program in BASIC at first (slow, indeed) and, after all bugs removal, simply rewrited that to assembler.
Although I use C in my job, I always try to avoid it and try to solve problems other way whenever it is possible. I have very bad experience about debugging huge programs in C. As I understood that is the main reason why is object language very popular (but still buggy).
I want to proove what I'm talking about; here is conquest to all who claim about speed against BASIC:
The task is:
1. Generate data base of at least 300000 points in space (X, Y and Z coordinates -- double precision float); use random numbers in desired area, e.g. x,y,z in range -3E9 up to 3E9 -- at least 12 significant digits
2. Create function which should find any point (if available) in given area:
function seekpoint(xmin,xmax,ymin,ymax,zmin,zmax)
3. Measure the worst case, i.e. time for finding particular point (or detecting that in given area is not defined any point)
You can:
========
1. Use any compiler or interpreter or CASE tool
2. Use any optimisation method
3. Use object language as well as assembler
4. Use any OS you want
Other limits:
=============
1. Database must exist on hdd
2. Database must have near unlimited capacity: at least billion points or up to disk capacity (data structure must accept at least billion point)
3. Direct reading from hdd is preferred, i.e. without use disk cache or RAM disk
If interested, just send me your results. Pls, send me your results using http://www.iridis.com/bps/contact.htm, or post to alt.lang.basic newsgroup.
Currently, I am preparing a database with 1000000 points for test the speed of my algorythm on my P5/166 computer. As soon as possible I'll post results in comp.lang.basic and on my site (see bellow).
If there should be enough interest, I'll put results on my site, of course with permission by authors.
Award
1. Winner should receive my source code written in QB4.5 :)
> We Come In Peace...Then We Create The War...
:))
Word disasters in 19xx:
Hiroshima '46
Chernobil '86
Windows '95
Croatian elections '97
- soon: programmers' nightmare '99
:)
--
Branimir Petrovic, Rijeka - Croatia
mailto:bps_ AT _iridis DOT com <- pls. remove underscores
http://www.iridis.com/bps
-------------------------------------------
Four steps of project:
1. Uncontrolled enthusiasm
2. Understanding that problem has no solution
3. Punishment of all employees under project
4. Award rest employees
- Raw text -