PASCAL as a system's programming language

[Thomas M. Breuel]

	The fact that PASCAL has clear semantics and consistent syntax
makes it very usable for 'Programming 101'. The fact that
PASCAL has also a rich set of data structures and rudimentary
mechanisms for modular program design allows it to be used for system's
programming as well.
	The language definition by Jensen&Wirth leaves extensions for
integration into a host environment unspecified (although an example is
given). All implementations that I have encountered (except for
Berkley/UN*X 'pc') provide such extensions, and with them PASCAL is an
excellent tool for system's programming. Calling it a 'toy' language
is completely unjustified.
	'C', on the other hand, is a usable, although a bit outdated,
workhorse for architectures very similar to the PDP, VAX, or
68000.  Unfortunately, 'C' can be adapted only with great difficulties
to other architectures, like the Cyber 173, the DEC-20, &c. (not to be
misunderstood, 'C' compilers for these machines exists, it is just that
practically no program that Joe Random Hacker produces on a VAX will
run on them unaltered). Other problems of the language include that its
syntax appears convoluted or counter-intuitive to many people (e.g.
"return (fun)(*(fun)rtoi(cdr(def)))();" ), and that 'C' is hard to
optimise or adapt to parallel execution.
	I don't want to start another PASCAL vs. 'C' debate here.
Which language you prefer depends mostly on your taste (both of them
are Turing equivalent :-).

					Thomas.
					breuel at harvard.