Re: Do college AP tests require C++ knowledge?

[clin@cs.umd.edu (Charles Lin)]

    You can get some of these answers, I suspect, by just heading
out to the Web page for the AP stuff (probably www.collegeboard.org
or www.ets.org), and they probably have a telephone number as well.
But, yes, I believe the tests are making the transition to C++.

    To address a previous poster's point about testing in a particular
language.   I'm not sure there are any easy ways around that.   Any
test that wants the reader to understand code is going to have to
be written in some language, even if it is pseudo-code.   They can
have the students write code in whatever language they are comfortable
with, but this would make grading a pain.

    The move to C++ underscores a few things happening in industry.
First, that the OO hype of the 80's has worked, and it's worked
primarily through C++, and now Java (though some will argue that
true OO languages are more like Smalltalk and Eiffel, and that
C++/Java are some bastardization of OO).    People seem to believe
that the OO style is the right one, or at least, industry believes
it (as opposed to, say, functional programming, which is perhaps
an academician's answer to a good programming language).

    It's true you can be taught concepts about programming languages
that are mostly independent of a language.   Ideas like assignment
statements, if-then-else, while loops, arrays, functions, pointers,
and even objects.    However, it's made much more clear if the
student can actually write programs in some language, and it makes
it easier for people to communicate if the language is commonly
used, such as C++.   In addition, the C++ exams would presumably
have more emphasis on OO design than a Pascal exam (although the
difference may be a little on OO vs. none).

    The sad part about software is that it is highly dependent
on language and OS, and the language is C++ (or possibly Java),
and the OS is Microsoft.   The reason for the choice of language
and OS has more to do with marketing and what everyone else
is using rather than which language or OS is best.   People won't
program in languages like, say, ML, because no one else is
programming in it, and so few people know the language or
can program well in it, so it's not considered a viable language
for solving real worled problesm, and this really has more to do 
with the popularity of a language vs. how good a job as a language
for solving problems.

    I guess the real issue is what should these tests test.  Part
of it is general programming principles, but in the real world,
that's always just going to be one part.   This will give you
some ability to move from one language to the next, but each
language not only has some syntactic peculiarities that might
make it difficult to learn, but it often has different features
or even a different paradigm.    The kinds of programs people
write in Java are probably going to look somewhat different from
the ones in C++, and that's not just because of the difference
in syntax, but the difference is features of the language.

    I believe that general programming skills only get you so
far in a programming language.   If I had a formula, I would say
that learning a programming language is a combination of knowing
general programming skills (modularity, loops, recursion, pointers,
etc) and then learning the details, strengths, and idioms of a
given programming language.    It's the second part that makes
a person a "power user" or "power programmer" of a language, rather
than someone who is trying to make their Lisp code look like C
since they know C.

--
Charles Lin
clin@cs.umd.edu