Re: Designing vs. Implementing, Was: Why context-free?

=?ISO-8859-1?Q?J=FCrgen_Kahrs?= <Juergen.Kahrs@vr-web.de> writes:
|> How can we change this ? What do I have to tell a skilled
|> implementor to make him change his mind when his design is bad ?
|> Whenever I try to do this, I run the risk of beeing treated
|> as a "bean-counter who simply doesnt understand".


Whereas I get told that I am an academic theoretician who doesn't
understand - despite my record in implementation and support, some of
which has been commercial!


I don't know. What I do know is that X amount of effort spent in
designing for implementation cleanliness, failure handling, error
detection and diagnosis saves 3-10 X effort in debugging and support,
and often speeds up the time to the first genuinely working product.
But it slows down the time to demonstration considerably, which is bad
for researchers who have no interest in anything beyond that and
developers controlled by "all that matters is time to market"
salesdroids.


The difference in providing facilities for end users and for
third-party developers is similar, but is a lot more subtle. I do
know that one of the major reasons that it is so hard to get the
messages across is that few people nowadays have any real familiarity
with all aspects of developing, supporting and using products. That
never was common, but is as rare as hen's teeth in people under 50.


Regards,
Nick Maclaren.

Return to the comp.compilers page.
Search the comp.compilers archives again.