🚀 go-pugleaf

Bear Giles (bear@tigger.cs.colorado.edu) wrote:
> In article <46giru$r6a@natasha.rmii.com>,
> Colin James III (The Rt Rev'd) <cjames@cec-services.com> wrote:
> >
> >How horrible:  all the positions are for C++ (with some Smalltalk?),
> >which means the resulting products shall be:
> >
> >1.  Unportable to many, many platforms;

> What on earth are you smoking?

> C and C++ are probably the most commonly available languages, and
> if the person writes POSIX compliant ANSI-C and draft ANSI-C++
> portability is _not_ a problem.  The same code should compile
> on damn near every Unix, DOS, Windows-NT, OS-2, et al system.

The problem is that in C and C++ there are too many things with "undefined"
semantics which can have "undefined effect" and the compilers do not,
or are not presently able to flag or forbid them. 

This property is not shared with quite a number of other languages.

So yes, if one writes perfect programs they will be portable, however
C and C++ technology does not always make it so easy compared to 
alternatives.  

Honestly I don't think that these 'holes' in C/C++ contribute as much
to importability as outright bugs. 

For instance, M3/Sather/Eiffel/Ada have very few to no implementation-defined
effects, and the ones that are there are localized into explicitly denoted
circumstances/libraries/classes/annexes. 

> >3.  Unmaintainable because C++ (and Smalltalk) is not inherently
> >strongly typed and hence allows legion side-effects;  and

> Yeah, right.  C++ is one of the strongest typed languages in
> common use.

If common use includes only C, Smalltalk and Visual Basic, then this
is correct.  But it is not correct compared to the kinds of technology
that CJ III espouses (no matter his bizzare style) and technology that has
been long available in research insitutions. 

> (Yes, there are cases where the equivalents of casts are required
> in Ada.  Take a close look at RW file I/O sometime.)

> I'll grant that casts can be abused, but competent management will
> prevent this from happening.  Once again the blame belongs on
> _management_, not the language, if the feature is abused.

I feel a purpose of technology is to reduce the unfortunate and malevolent 
consequences of human frailty.

> Sorry for having a C++ discussion on sci.crypt, but these statement
> are so outrageous that they can't go uncommented.  In fact, many of
> us would have serious doubts about any new software effort which
> _didn't_ use C++.

oops, have I been trolled?

> Bear Giles                     For every complex problem there is an
> bear@cs.colorado.edu           answer that is clear, simple and wrong.
> copyright 1995                                         -- H L Mencken

cheers
Matt