Re: "Size" of Ada vs. C++

[Richard Riehle <richard@adaworks.com>]

Florian Weimer wrote:

> Anyway, I think it's rather strange to judge a language by the "size"
> of its definition.

Absolutely correct.  I have seen people compare languages based on the
number of reserved words (less is better),  the size of a "hello world"
program's executable (less is better),  the number of predefined libraries
(more is better),  the number of people who use it for programming (more
is better), the number of ads in the help wanted section (more is better),
and on and on and on.    There is no end to the silliness that people use
to compare programming languages.   The Modula-3 exercise that attempts
to define a language in some number of pages of the reference manual also
approaches absurdity.

A programming language needs to be judged on its expressiveness first.  One
factor in this judgement is, how well it expresses the software problems it
is intended to solve.  We could reframe this as, how well does the solution
space (represented by the language and its corresponding tools) map to
the problem space?   Another factor derives from Donald Knuth's notion of
"Literate Programming."   This notion suggests the question, how well
does the language support the human process of developing software?
There are other factors, but this margin is too small to enumerated them all.

In evaluating a programming language, we need to define qualitative
criteria as well as quantitative criteria.    Sadly, most of quantitative
criteria are flawed due to the immature reasoning that leads to their
selection.   Sadder still, such reasoning is likely to prevail for a very
long time, given the current state of software engineering practice.

Sorry for my pessimism.

Richard Riehle