Re: ADA and Pascal SUCK, C,C++, and Java are the only languages you need!!

[dewar@merv.cs.nyu.edu (Robert Dewar)]

John Stevens <jstevens@samoyed.ftc.nrcs.usda.gov> wrote:

>Well, considering I was unable to port an Ada program to two of the five
>platforms I was targeting, while I was able to port a C program to all five,
>I'd have to say that C is infinitely more portable.

No, you would have to say that C was 5/3 times as portable in your scenario. :)

>YMMV, and all that.
>
>>If you have an Ada compiler
>
>That *IF*, being the gotcha.
>
>Has anybody ever successfully used Ada on a Linux box?


Robert Dewar replies

Since you are clearly unaware of what Ada compilers are available, it is
quite possible that your failure to successfully port Ada to some of your
platforms was simply a matter of not finding out how it could be done.
Yes, of course people have used Ada on a Linux box. As you can easily
discover if you look just a little bit, GNAT is fully implemented and
fully validated on Linux, as well as a dozen other platforms. We have a
number of commercially supported customers using Linux.

You can incidentally immediately discover that GNAT is fully implemented
on Linux by accessinfg the official list of validated Ada compilers. Any
Ada programmer should be aware of how to do this, since this is all publicly
available information.

The implementation of Ada 95 on Linux, like all other implementations of
GNAT supported by Ada Core Technologies, is a 100% full implementation of
the core language and all the annexes. We regard it as particularly
significant, since it means that GNAT + GNU + Linux form a 100% free
software solution in the Ada 95 area.

There are many reasons for difficulties in porting programs in any
languages:

  o incompetent use of the language
  o legitimate implementation dependent features
  o inadequacies in compilers
  o fundamental portability flaws in the language design

In the case of Ada, with the set of implementations available today, there
are very few portability problems if any that can be attributed to the
last two reasons. Incompetent programmers can of course write non-portable
code in any language. In particular, to write portable code you MUST be
aware of the defining standard, it is not good enough to informally knows
what happens to work with your compiler.

One real advantage of Ada over C is that Ada programmers tend to be at
least somewhat familiar with the Ada standard, and can be expected to
have a copy available, at least for reference purposes. The C standard
is very much less well known among C programmers (you only have to look
at some of the idiotic statements posted about C in the sizeof(int)
discussion to see how many C programmers have strange ideas about C).
Alternatively ask a roomful of Ada programmers how many have a copy of
the standard available to them and routinely reference it. You will
get essentially 100% yes response. Now ask the same question to a
roomfull of C programmers.

One reason for the difference is that the Ada standard has always been
completely freely available electronically, without copyright problems,
and inexpensive printed copies have also been available. The Ada community
fought hard to achieve this (at one point the ISO standardization was in
the balance over a resulting copyright controversy). Many other language
standards have not been so fortunate. I don't know the current situation
(I know that Ada has inspired other standards groups to try to liberate
their standards), but certainly earlier, the ANSI C standard was relatively
unavailable, and expensive.

Another reason is the culture and circumstances. People tend to have learned
C informally, and figure they know it without need to reference a standard.
Indeed I often meet C programmers who don't really know what the ANSI
standard for C *is*, let alone having a copy of it that they have read.
In the Ada world, everyone knows about the Ada standard, and is very aware
that knowing the standard and following it carefully is the key to writing
portable code.

Robert Dewar
Ada Core Technologies

P.S. Of course I am quite aware that substantive technical arguments can
be made regarding the superior portability of Ada over C from a language
point of view, but those arguments are well covered, and familiar. I wanted
to focus on a rather different aspect of portability, which is the role of
programmer attitude and knowledge.