🚀 go-pugleaf

In article <45e8r1$9sr@shiva.usa.net>,
Siegfried Heintze  <72740.1365@compuserve.com> wrote:

>I not sure I feel good about the people at Columbia (or whereever) 
>encouraging students to learn C++, but we have no control over that. The 
>fact is that by 1997 there will be an advanced placement exam in C++ and I 
>suspect very, very few high school teachers know C++ let alone C.

Well, maybe.  If industry or academic organizations, which surely many
people around here are involved with (not me, though), really decided
it was a bad idea, they could apply at least some sort of pressure to
the AP exam system.  Might not be effective, but I wouldn't go so far
as to say we have *no* control over it.

>Or does anybody believe that our best contribution would be to discourage 
>learning C++ until they are at the University?

People taking this approach underestimate highschool students, I
think.  I'm rather past highschool now, but in 1968 and 1969 I taught
myself Fortran and IBM 1620 assembler and IBM 1401 assembler out of
books.  I was in 8th and 9th grade at the time.  At least half a dozen
of my friends taught themselves Fortran or BASIC at about that time,
and some went on to assembler.

AP courses and tests aren't for the basics and for all students;
they're for college-bound students who have taken special courses in
an area.  Those kids, at least, should be perfectly capable of
learning C++.

Teachers to teach it are more of a problem -- but again, we're talking
about AP courses, not your run-of-the-mill everybody-takes-it course,
so they don't need as many.

I don't think C++, taught by a decent teacher, is harder than
assembler self-taught.  And learning program design object-oriented
from the start might work out pretty well.
-- 
David Dyer-Bennet	  ddb@network.com, ddb@terrabit.mn.org, ddb@gw.ddb.com
SF cons: http://www.ddb.com/4th-Street, http://www.mnstf.org/minicon31
Me: http://www.ddb.com/~ddb (photos, Olympus photo eqpt. for sale, sf)

Siegfried Heintze <72740.1365@compuserve.com> wrote:
[snip]
>I not sure I feel good about the people at Columbia (or whereever) 
>encouraging students to learn C++, but we have no control over that. The 
>fact is that by 1997 there will be an advanced placement exam in C++ and I 
>suspect very, very few high school teachers know C++ let alone C.
[snip]
>
>               Thanks,
>                      Sieg
>

An AP test in C++?  That sounds so strange as to be ridiculous.  I can't
imagine an AP test in any specific language.  AP tests are for gaining
college credit for high school courses based on the AP test score.
C++ is not a course in any CS department I've had exposure to.  There
are programming courses taught using C++, C, Pascal, etc, and object oriented
programming courses taught using C++, etc.  I can imagine an AP test for
object oriented programming, or in basic computer science, but in C++?

--Paul L.

Why would anyone ever do a bubble sort, anyway?

Jim Shapiro

In article <45jlpi$3ki@CUBoulder.Colorado.EDU>,
shapirj@rintintin.Colorado.EDU (Jim Shapiro) wrote:

> Why would anyone ever do a bubble sort, anyway?
> 
> Jim Shapiro

In order to understand why to study other types of sorts.  I teach
programming to high school students and feel that it is necessary to
explore a wide range of algorithms in order to understand their
limitations and advantages.  

Vince Devlin

In article <nelsoni.813597537@rintintin.Colorado.EDU>,
Ian S. Nelson <nelsoni@rintintin.Colorado.EDU> wrote:
>shapirj@rintintin.Colorado.EDU (Jim Shapiro) writes:
>
>>Why would anyone ever do a bubble sort, anyway?
>
>>Jim Shapiro
>
>There are tons of applications where bubble sort is one of the best algorithms.
>Various sort algorithms can be speed up a lot by not completely sorting 
a list.
> [ rambling deleted ]
>Sorry to ramble, but I can't stand by and let somebody knock (even in jest)
>everybody's favorite sorter...  It's an amazingly useful algorithm.

From 'Numerical Recipes - The Art of Scientific Computing' by Press et al
Page 227:

"We _will_ draw the line, however, at the inefficient N squared algorithm
_bubble sort_.  If you know what bubble sort is, wipe it from your mind; 
if you don't know, make a point of never finding out!" (Emphasis not mine.)

Sorry to bust your "bubble".

Regards,
Jim Shapiro - number theorist and Perl pragmatist

shapirj@rintintin.Colorado.EDU (Jim Shapiro) writes:

>In article <45mjvk$pil@CUBoulder.Colorado.EDU>,
>Robert Tarrall <tarrall@plutus.Colorado.EDU> wrote:
>>Jim Shapiro (shapirj@rintintin.Colorado.EDU) wrote:
>>: Ian S. Nelson <nelsoni@rintintin.Colorado.EDU> wrote:
>>: >There are tons of applications where bubble sort is one of the best algorithms.
>>: >Various sort algorithms can be speed up a lot by not completely sorting 
>>: > [ rambling deleted ]
>>: >Sorry to ramble, but I can't stand by and let somebody knock (even in jest)
>>: >everybody's favorite sorter...  It's an amazingly useful algorithm.
>>: 
>>: From 'Numerical Recipes - The Art of Scientific Computing' by Press et al
>>: Page 227:
>>: 
>>: "We _will_ draw the line, however, at the inefficient N squared algorithm
>>: _bubble sort_.  If you know what bubble sort is, wipe it from your mind; 
>>: if you don't know, make a point of never finding out!" (Emphasis not mine.)
>>: Jim Shapiro - number theorist and Perl pragmatist
>>
>>And if it's in NumRec, it's gotta be right, eh?

>I think I would say yes to that question.  I was the original alpha 
>tester for the FORTRAN version of "Recipes" and it was amazingly solid 
>for an alpha.  In 200+ functions I located exactly two bugs.  That is not 
>to say that I agree with them everywhere, but it is a _very_ good 
>reference.  Can you tell me any place (besides this one) where you have 
>found them to be wrong?

So they say to forget about bubble sort and they are "correct?" That is a very
subjective thing to say, can they prove their correctness?  
I studied CS at Carnegie Mellon University with Merrick Furst, Danny Sleator, 
Jim Morris, Avrim Blum, Guy Steele and numerous other well respected scientists
and some of them agree that bubble sort is near useless, some see it as a great
tool in some places.  Who's to say?  

>I do not what languages you use, but I no longer code _any_ sorting 
>algorithms any more than I code sine and cosine.  I have done both, but at 
>least in C (qsort) and Perl (sort) sort is built in and at the system 
>level the sort routine is extremely fast, flexible and solid.

It's not a stable sorting algorithm (the C stdlib qsort.) 
What do you do if you need a stable sorter?  Or what does Num Rec. suggest for
stable sorting? 

 I don't really code sorters anymore either, I wrote a nice C++
templated sorting object with all of my favorite algorithms (actually,
all the algorithms I knew of) and I just use it and decide which
algorithm to use for whatever task I'm doing.  I do have to code in
other languages fairly often and I just cut and paste and do a quick
port most of the time.. Only SML has given me any grief...

matt@godzilla.EECS.Berkeley.EDU (Matt Austern) writes:

>In article <45mjvk$pil@CUBoulder.Colorado.EDU> tarrall@plutus.Colorado.EDU (Robert Tarrall) writes:

>> I'd side with Ian on this one.  You deleted what I think was his
>> most relevant point, which is that a bubble sort is incredibly
>> easy to code.  If you don't need to sort more than 100 or so items,
>> it's not in a time-critical section of code, or you're just writing
>> this for yourself, why not use a bubble sort?

>Because there are other sort algorithms that perform far better (are
>you certain that program that you initially thought would just handle
>a hundred items will never be modified or used in different contexts?)
>and that are almost as easy to implement.  A good quicksort
>implementation is tricky, but there's nothing hard about heapsorts or
>Shell sorts.

It's usually a safe guess that it will never be expanded.  From a quality 
perspective, if somebody is messing with my code in the future, he/she should
pay attention to all of it. That way when they break the 100 element threshold
they can plug in a new sorter which is a trivial task. 

>It'll take slightly more time for you to write a routine that does a
>Shell sort than one that does a bubble sort, but only slightly.  And,
>of course, you shouldn't be writing sort routines all that often
>anyway!  Ideally you ought to use one that someone else already wrote
>for you, like qsort() from the standard C library or sort() from the
>STL.

Yep..  There are applications where heapsort and quicksort are not acceptable
though.