I think that industry has very different viewpoint on usability than
does the research community.  (In case anyone asks, I currently work on
developing a product; the development is funded by a for-profit company
and will be sold by the same for-profit company.  Any research that gets
done is entirely by accident.)

For us, usability equals productivity.

Suppose that someone offered us a wonderful new language, designed to
be a big improvement when evaluated using the criterion that Thomas
has developed.  Let me further fantasize that a mature implementation
is available for our target platform  (big fantasy!!)

Would I advocate using this new language for our project?

No.  Not even if you guarenteed that we would never again make coding
bugs using the new language!!

Why not?  Well, most of the problems we encounter, the really costly ones,
could not be solved, or even impacted, by changes at the programming
language level.  They are at the level of requirements/design.
(Anybody know a good way of controlling user access to objects stored
in a relational database?)

Suppose that someone offered me an improved version of a UML tool that
did a better job of handling constraints, but it only could generate
COBOL output.  I'd have just two questions:  (1) Where do I send the
check and (2) what version of COBOL do I need to get?

In the world of modern industrial software development, improving the
design of programming languages - systems at the level of Python/Java/C++
is just about as useful as improving the naming of op codes for assembly
language.

[Another message will follow on one industrial viewpoint on Java]

Ruven Brooks

Reply via email to