Re: Is C/C++ the future?

[riehler@ajpo.sei.cmu.edu (Richard Riehle)]

In article <1994Oct30.210203.1863@muvms6> hathawa2@muvms6.wvnet.edu (Mark S. Hathaway) writes:

>If you don't think Ada is complete, but Fortran is then I'm confused.

  Ada is just as complete as FORTRAN.  The issue is platform targeting.
  

>Isn't the problem of libraries one of major importance these days?  Every
>language has it's own libraries and they're not compatible and some are
>so closely attached to an operating system that they're not very portable
>and...<argh>

  I would agree that the issue of incompatibility has been a serious 
  problem with Ada 83/87.  Each compiler vendor took a different approach,
  if any, to defining libraries for targeted platforms.  Often, early Ada
  compilers lacked any useful libraries.  This improved in the late eighties
  and early ninties.  

  Ada 9X defines uniform package specifications as a remedy to this problem.


>Once you've got a good procedural language like Pascal, Modula-2, Ada, C,
>etc. there should be an extension of that utility by having better
>reusability across languages.

 None of the languages you name have fully compatibile libraries.  C, as a
 universal assembler, includes good support for hadware level programming,
 as one would expect of any other assembler.  The other languages are 
 high-order languages that require system-level libraries.  These are not
 defined as part of the language design.  
 
>What if a new language was specifically designed to allow for the use of
>libraries written in COBOL, Fortran, Pascal, Modula-2, Ada, C, Smalltalk,
>Lisp (or whatever widely-used languages could be included)?

 Interesting notion. Have you been peeking at the Ada 9X reference manual?

>Then, what set of functions/procedures/classes/etc. would constitute
>completeness?

 Each hardware and operating system environment is unique.  To get the best
 performance from such environments will always require libraries that take
 advantage of the best of the unique features.  Sometimes, the absence of
 such libraries will require interfacing with other langauges, machine code
 insertions, or whatever ...  

 This is not sinful and evil. It is not even situational ethics.  Rather,
 it is a reality of using high-order languages where appropriate, and 
 low-level langauges such as C and assembler when appropriate. 
>
>BOO,

  HAPpy Halloween to you, too.  :-)


   Richard Riehle