[comp.lang.misc] Languages for NumAnal: my opinions

ajayshah@alhena.usc.edu (Ajay Shah) (04/03/91)

Fortran:
	- great base of library code around
	- your boss/advisor probably uses it
	- you don't want to touch it.

Pascal:
	- Standard pascal is crippled.
	- Turbo Pascal is fantastic, but locks you into the PC,
	- strong strong typing cripples scientific libraries (OOP has
		solutions, but that locks you into the PC even more).
	- you don't want to use the PC!

C:
	- Better than Fortran as a language,
	- decent libraries available, especially with f2c around,
	- terrible syntactic&semantic irritations like clumsy matrices.
		CS types may claim C's pointers are elegant, but to a
		mathematician they're a pain in the butt.
	- great portability, the best available today.
	- excellent hooks into Unix internals; standard C library is
		wonderful -- so everytime you need to do something
		non-number-crunching you are well off.

C++:
	- Gorgeous code quality enabled by operator and function
		overloading
	- Usual C strengths on portability-Unix integration, availability
		of old fortran libraries via f2c.

Modula, APL, lisp etc:
	- Way too flaky for serious consideration.  
	- I love what I've read about Modula but I don't want to
		touch something so far from being a standard.


In all, I'm moving to C++ for my numerical work.

-- 
_______________________________________________________________________________
Ajay Shah, (213)734-3930, ajayshah@usc.edu
                             The more things change, the more they stay insane.
_______________________________________________________________________________