My childhood dream was to become a university professor in theoretical computer science or maths... Yah I know... I was a weird seven year old!!!
Well I'm still working on it!!!
As for the best starting language thing.
I have never met a better programmer than one that has followed this path:
Relatively formal foundation in mathematics -> Functional language that's strongly typed (e.g. Haskell, my personal favorite) -> Study of algorithmics and formal logic -> Functional language, preferable OOP (there I completly agree with C++) -> Some study of asm.
The relatively formal foundation in mathematics is important I think as it teaches one the correct methods of thinking and makes it easy to learn the syntax and workings of a functional language (haskell is again best syntax wise for crossing the math -> functional language boundary).
It is also important as the two disciplines are often strongly interlinked and it gives a good intelectual foundation for the study of some of the harder concepts and problems one will grappel with later.
Learning a strongly typed functional language will encourage good coding practices, no fudging around with type casts.
These types of language also make the coding less important, one tends to solidly plan and design an algorithm by breaking down the problem into the smaller sub problems that compose it, as opposed to jumping straight into the code with only a vague plan and fudge around until it works.
It is also important as the simplicity of the languages syntax allows one to really focus more on the problems being solved.
One really cannot compare:
Code: Select all
#include <vector>
....
vector<int> x;
for(int n = 0; n <= 10; n++) x.push_back(n);
...
with the simplicity and ease of comprehension of:
These points as well fact that many of them also have a large set of integrated set of functions and integrate bignum support makes them the ideal play thing for a budding programmer with the correct foundation, as they remove the need for playing around with bignum, etc.. libraries and associated problems for a begginer.
Now don't get me wrong, strongly typed languages, although they have the advantage that one has to do something fairly radical to create something that will compile and then cause a segfaults, etc... They are not the ideal languages for real world development, although are far better for theoretical studies, so I love them!
Next, I think it's important to study algorithmics to learn about new paradigms as well as new ideas.
Many problems can be shown to be equivalent to other well studied problems, so a study of these main, well studied problems (i.e. tree searches, 10 knapsack optimisation problems, etc...) is extremely important.
A foundations in formal logic is important as:
a) It further encourages the right way of thinking and gives a good understanding of how the programming works, as well as showing one some interesting aspects of computation, such as it's limits (Gödel's incompleteness theorems).
b) It allows real debugging, not testing it with a set of potentially tricky as well as other more trivial inputs and checking if the correct answer is outputed, but writting the program as a functional statement, converting it into either a statement in lambda calculus or mathematical logic and finding a real proof that the program will always give the correct answer to the problem (given certain assumptions such as infinite memory and time) within certain constraints.
Next, for real world applications, a good OOP such as C++ should be learnt because of it's wide array of libraries, the fact that OOP is an ideal language for representing certain real world problems easily and intuitively and the fact that it is an imperative language, therefore makes it (mostly) ideal for real world applications.
Finally, one should learn asm, as it gives one a look at what is 'under the hood', which can be quite interesting! Also, it really helps with debugging and understanding certain errors. One could argue that the culmination of this would be to do at least a little OSdev!
I really think this would make for the best "real world" programmer!
Just my two cents, sorry, I know it's very long and also very off topic, but hey, I'm very tired, so I only realised after I'd wrote it, so I'm going to post anyway! (Sorry admins, I hope you won't have to move this!)
Kind regards,
Jules
P.S. I'd also advocate a study of basic complexity theory somewhere along the line, it can be very interesting and useful!
P.P.S Also, haven't been around for some time, so hey guys!
I remember trying to draw it in paint on windows 95.
)