Re: 30 Years

[cis.ohio-state.edu!magnus.acs.ohio-state.edu!math.ohio-state.edu!howland. reston.ans.net!noc.near.net!inmet!spock!stt@ucbvax.Berkeley.EDU (Tucker Taft)]

In article <EACHUS.93Sep10152513@spectre.mitre.org> 
  eachus@spectre.mitre.org (Robert I. Eachus) writes:

>In article <26qc0u$k0b@louie.udel.edu> 
  carroll@gloin.cis.udel.edu (Mark C. Carroll) writes:

>  > This is made even worse by the way in which Ada documents describe the
>  > language. I got the annotated reference manual for Ada9x, and went to
>  > print it out. How long could it possibly be? I use languages similar
>  > to Ada all the time, and the manuals are between 30 and 100 pages
>  > long. The Ada manual is over *500* pages, the overwhelming majority of
>  > which is bureaucratic twaddle. 
>
>     How many programming language standards have you read?  The
>Pascal standard is small and readable, but, for example, the Algol 68
>standard probably wins all the prizes for obscurity and
>impenetrability, the size of the PL/1 standard (not PL/1 subset G)
>makes the Ada standard look like light reading, and (not to ignore
>popular languages) the COBOL RM easily surpasses even the AARM in
>bureaucratic twaddle (and size for that matter).
>
>     Language standards are usually designed to be read only by
>compiler implementors.  Ada broke that tradition by having a reference
>manual that really was laid out for users of the language to
>reference.  But it was certainly not intended to be an introductory
>textbook.  In Ada 9X, the RM is still intended for users, and the AARM
>is a bonus document (replacing the Ada 83 Implementors Guide) to help
>compiler writers.
> . . .
>     There was a lot of discussion on this subject (the apparent
>complexity of the standard) at the August meeting, and some progress
>was made, but if you have any ideas about how to make the presentation
>less daunting, let Tucker know.

Yes, please do as Robert Eachus suggests.  Furthermore,
we have recently decided to stop encouraging the "general"
public to read the Annotated Ada 9X Reference Manual (AARM),
exactly because of the problem you cite.  It has so much
stuff in it that it clearly detracts from overall understanding.
The AARM will continue to be available to the public, but we
will emphasize that the unannotated version will be the official
standard, and is the only one explicitly intended to be reasonably
accessible to the average programmer.  Even better, hopefully,
will be the Rationale document, a new version of which will
be released shortly.

Perhaps the surprising thing about the Ada standard is that
average Ada programmers actually do read it, whereas average
Cobol or Fortran programmers rarely read the corresponding
Cobol or Fortran ISO standards.

It is a real challenge trying to address the requirements for
precision and completeness that is appropriate to an
international standard, while addressing the requirements
of ease of reading and understanding that is appropriate to
a manual that will be read by typical programmers.

The Ada 9X reference manual has probably swung a bit
too much over to precision and completeness and away
from understandability, and we hope to swing it back a 
bit during the public review process.  Please let us know
if you find clauses that are generally mystifying or full
of unnecessary "bureaucratic twaddle" (B.T.).  We will do our
best to slim them down, without abandoning completeness
and precision completely.  The typical solution will be
to move the B.T. to the AARM, so don't expect that document
to become any easier to read.  The unnannotated reference
manual, however, will hopefully become easier to read
as a result of the review.  

As someone great once said, "If we had more time 
we would have made it shorter..."


>					Robert I. Eachus

S. Tucker Taft    stt@inmet.com
Ada 9X Mapping/Revision Team
Intermetrics, Inc.
Cambridge, MA  02138