[net.lang] Language level measurements query

bob@sdcsvax.UUCP (Robert Hofkin) (02/02/85)

We've been talking a lot of folklore and hypothesis about "which
languages are higher level".  I think most of us have found one or
another programming language to be more productive for us (for certain
classes of problem, anyway).  But personal bias ("it works for me, so
it must work for you) isn't very convincing.

Does anyone have decent MEASURES of language "level," one against
another, on a set of programs?  To me, that would be the amount of code
AND length of time a programmer competent in the particular language
would need to write those programs.  Average lines of code per day is
insufficient!

If so, please send me references.  If not, I have an idea for a USENET
experiment.

--Bob Hofkin   ...!sdcsvax!bob

stewart@houxf.UUCP (Bill Stewart HO 4K-435 x0705) (02/11/85)

The problem with even proposing level measurements is that "level" isn't a
one-dimensional variable - some of the things that go into the concept are:

	What kind of "high-level" abstract objects are available
	What "high-level" verbs can you apply to those objects
		(or, for object-oriented languages, what actions can you ask
		the objects to do for you.)
	What kind of control structures does the language provide
		this definition applies more to algorithmic languages
		than to functional ones.
	What are the "lowest-level" objects you can manipulate (low-level in
		the sense of machine-dependent, implementation-dependent,
		"real stuff".)
	What are the lowest-level objects you *have to* manipulate to do
		useful work?  (In APL you can use flexible-sized arrays; in C
		you need pointers and often malloc().)

The latter two are easy to mix up; people like Jerry Pournelle think BASIC
is higher level than C because pointers let you know about objects'
locations in memory; I guess BASIC is low-level if you include PEEK and POKE
and high-level without them.

Also, a language can be "higher-level" for some tasks than for others; I'd
rather write simulations in C++ or LISP,  but I'd rather crunch statistics
in APL.

-- 
Bill Stewart	ho95c!wcs AT&T Bell Labs, Holmdel NJ
HO 4K-435 x0705   (201-949-0705)
{allegra, ucbvax!ihnp4, decvax!harpo}!houxf!stewart
------
Sorry if the articles I'm replying to re ancient; we lost news for a month.