Re: Ada 9X Mapping

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

>     I have not seen any discussion on the Ada 9X draft Mapping Document in
>this group.  Is there another group dedicated to informal 9X rapping as opposed
>to the formal comments to the AJPO?

>Or does everyone thinks it perfect? 8-)

Well, personally I think it's overkill. The current documents are in 
a very sketchy state, and they are already large. By the time they
are transmuted into real modifications to the LRM, the LRM will be
twice as large as it already is.

I'm personally a big fan
of subprogram and package types. I also favor getting rid of the
idiot special-cases (e.g. latter declarative items, etc) that make
learning the language more difficult than it needs to be. And I'd
like a few notational conveniences like "return...when" and "raise...
when".

The problem is, MY list isn't necessarily the same as YOUR list: and
our aggregate lists may not include some other person's wish list.
The aggregation of EVERYBODY'S lists results in a huge shaggy baggy
monster of a language revision. What should have been a very quick,
very restricted effort to triage the top 20 complaints has mushroomed
into accomodating just about everybody.

The question we ought to ask ourselves at this point is: what problem
are we really trying to solve? Are we trying to fix some defects in
the original language definition? Fine--by all mean let's do so. Or
are we actually trying to permute Ada into some radically new language,
like, say, C++ or Eiffel? If so, WHY? The putative merits of such
languages for large complex systems are not well-established. I think
they're largely a fad. Do we run off in eleventy-seven different directions
trying to make Ada be all things to all people (something it was never
EVER intended to do), or do we stick with the knitting? Focusing on
finding out WHY many compiler vendors have crappy tasking applications
is probably a far better way to deal with performance issues than is
dragging in a whole new, untried mechanism. If we're going to make the
language be all things to all people, shouldn't we throw in inferencing
for the AI folks, multiple inheritance for the inheritance weenies,
dynamic typing for the terminally confused, etc etc etc?

Hrmph.
--
* The opinions expressed herein are my own, except in the realm of software *
* engineering, in which case I borrowed them from incredibly smart people.  *
*                                                                           *
* Rational: cutting-edge software engineering technology and services.      *