Re: C/C++ knocks the crap out of Ada

[adaworks@netcom.com (AdaWorks)]

Kevin Cline (kcline@sun132.spd.dsccc.com) wrote:
: In article <adaworksDoBsy8.Brz@netcom.com>,
: AdaWorks <adaworks@netcom.com> wrote:
:  >Kevin Cline (kcline@sun152.spd.dsccc.com) wrote:
:  >
:  >

: I suppose every language design error could be classified as an
: inconvenience, since there is almost always some workaround available.
: But the following missing features in Ada-83 were serious problems
: when developing hosted applications and directly led to the rejection
: of Ada by the marketplace:
:  1. Inability to pass a function or procedure as an argument.
:     This went far beyond an "inconvenience" for those of us attempting
:     to use event-driven GUI libraries.  There was no portable 
:     work-around for this problem.

   OK, Kevin,

   I'll give you this one. There were some non-portable workarounds, but
   this was a shortcoming of Ada 83.  The new ISO Ada standard supports 
   the ability to pass a function of procedure as an argument.

   However, for the next one:

:  2. No standard interface to any OS facility more advanced
:     than line-at-a-time input/output.  Also very difficult to
:     work around, particularly if trying to produce a portable program. 

   The only thing wrong with this was including it in the "core" of the
   Ada 83 definition.  Text_IO was never intended to be more than it 
   was, a universal scrolling, left-to-right and downward I/O package.
   The issue with all the I/O packages was portability.  It still is.

   Nothing in the Ada 83 design precludes the creation of I/O packages
   for other terminals, operating systems, and I/O devices.  In fact, such
   packages abound.  How do you think people use Ada for the huge range
   of operating systems on which applications have been deployed?

   The Ada predefined I/O packages were not intended to be platform-specific.
   The language designers expected compiler vendors to provide packages for
   I/O on their targeted platforms.  Many compiler publishers fell short in
   this area. This was especially true of "checkbox" compilers supplied
   by some of the hardware vendors. 

   Once again, the ISO 1995 Ada standard updates the facilities of the I/O
   packages, but portability is still a concern.  Therefore, interfaces to
   specific operating systems is outside the scope of the core language,
   as it should be.

   Richard Riehle
   adaworks@netcom.com
  
       BTW, a "checkbox" compiler is one which satisfies the minimum
       standard for validation so the vendor can check the box on the
       procurement request for validated Ada.  This practice created
       additional problems for Ada since the hardware vendor was only
       interested in getting a collection of computers in the door, not
       in seriously supporting Ada.  To avoid lawsuits, I'll not post
       the names of such vendors, but they know who they are.

       RR
-- 

richard@adaworks.com
AdaWorks Software Engineering
Suite 27
2555 Park Boulevard
Palo Alto, CA 94306
(415) 328-1815
FAX  328-1112