Re: Ada9x Transition Plan (443 lines) ftp-ed from AJPO

[jls@rutabaga.Rational.COM (Jim Showalter)]

>I couldn't agree more with your basic idea. However, industry folks should
>realize that in the university world we don't teach languages, we teach 
>concepts and the languages are just means to the end (this is especially
>true at the undergraduate level). Just because we teach Ada using spiraled
>subsets of the language doesn't mean we only teach "sort-of" Ada. 

I think we're in violent agreement here, actually. When I teach Ada, I
also teach using a spiral approach (what other method IS there?). I just
don't want to see subset Ada blessed by the validation office. If you
want to teach a subset, fine: but do it with a real Ada compiler.

>I keep
>meeting industry guys who are so far removed from what we do that they
>sincerely believe we can start with a first semester freshman and teach
>all of Ada in one semester. They oughta go back and imagine themselves
>before they ever write a program.

I am currently teaching a bright but scarcely computer-literate individual
Ada as his first programming language. We are currently a month into the
experiment, and he is quite capable with control structures, subprograms,
types, subtypes, private types, selectors/constructors/iterators, and
generics. I'll admit that sometimes I get asked interesting questions
(like, try explaining sometime why a formal is CALLED a formal...), but
all in all I think the experiment has been a resounding success--and we've
got two months left.

>The biggest problem I have with most Ada83 systems is that they were
>designed for industry, not for education. As such, the payoff from them
>comes with large(r) projects, not the little ones freshmen do.

Agreed, but only partially. I think any software engineering student
benefits from the orthogonal (well, MOSTLY orthogonal) syntax/semantics
of the language, the type model, spec/body separation, and private types:
with these you can attack 90% of the language-caused problems that beset
projects of any size.

>The
>nice thing about Arcturus is that performance, within its capacity limits,
>is directly proportional to program size. Perceived performance of
>commercial compilers is, for student-sized projects, nearly O(1),
>with a much-too-high constant.

We have an effective batch compilation rate of 150KSLOC/minute. Is that
fast enough? :-) [gotta LOVE incremental compilation]

>My recurring nightmare is that the
>government will end up funding another ALS or AIE which, by the time the
>cautious bureaucrats allow its release, will have long since been overtaken
>by events. RAPID PROTOTYPING is what this oughta be about.

Violent agreement here! ALS = infinite resource sink
--
***** DISCLAIMER: The opinions expressed herein are my own. Duh. Like you'd
ever be able to find a company (or, for that matter, very many people) with
opinions like mine. 
              -- "When I want your opinion, I'll read it in your entrails."