Re: Cost of Ada (Was: How to Make Ada more widely used?)

[pipex!warwick!zaphod.crihan.fr!univ-lyon1.fr!scsing.switch.ch!epflnews!di suns2.epfl.ch!lglsun!nebbe@uunet.uu.net (Robb Nebbe)]

In article <1vnjv8$9e5@spcot.sanders.com>, lewin@sanders.com (Stu Lewin) writes
:
: ...
: For example, there are tools available now that provide most of the Ada
: type checking at compile time and range checking at run time. We had a
: policy that all C code devloped needed to compile without error under
: four compilers with full warnings and ANSI compatibility enabled: "Sun"
: gcc, SunPro cc, DEC cc and HP cc.  Code was required to pass through
: lint (although it was impossible to get lint to stop complaining about
: some valid things). In addition, the code needed to load and execute
: without warnings in CodeCenter. We also used Sentinel and Purify to
: check for memory leaks and illegal accesses.

This could be automated with scripts and things but as far as simplicity
goes Ada wins hands down compared to this.

The question of whether the resulting code is of the same quality is
debatable but that isn't really the question. For me --but then I'm an
idealist-- the question is "what is the right way to do it?" You will
never convince me that using 4 compilers, lint, CodeCenter, Sentinel and
Purify is the right way to do it. I'm willing to admit that it is one
way, but there is certainly a better way. (Ada being one example)

(I must confess that I do more or less the same thing when I program
 in C so don't take this as a criticism of your development process
 but rather as a criticism that one must do this if they program in C. )

: ...
:
: But I do agree that the software development process has way more to do
: with writing good, safe, economical code than the choice of the
: language.
:

This is the equivalent of saying that object-oriented programming has
way more to do with encapsulating types and their operations than the
choice of a language. :-) (draw your own conclusions)


Robb