Re: Papers saying Ada as an overly complex language and hard to implement

[Brian Rogoff <bpr@shell5.ba.best.com>]

On Mon, 16 Feb 1998, Yongxiang Gao wrote:
> Robert Dewar wrote:
> 
> >  These five aspects are not only different, but mutually incompatible. For
> > example, these days, it is generally perceived that languages need to
> > be fairly feature rich (Ada 95, C++, Fortran 90, OO COBOL, ...) to be
> > simple to use. Even Java is pretty feature rich, especially if you include
> > its standard libraries.
> 
> For a beginner, "rich" comes with "complex to use".

Its quite easy to learn to use a subset of Ada which isn't that much more 
complex than Pascal. Indeed I subset every software tool I learn. So I
would argue that your statement is not accurate. It would be accurate if 
you needed to learn all of Ada's features and how they interacted before 
you did anything at all, but that isn't the case. That doesn't mean that I 
can't imagine a simpler beginner's language, like Scheme, just that your
statement above is plainly false.

> >  As for "hard to write a compiler for", Ada is no more difficult than
> > any of these other languages. These days the really hard part of any compiler
> > is doing a good job of optimizing the object code on modern
> > architectures, and this is about the same effort for any language.
> 
> Who tells you "Ada is no more difficult than any of these other languages"?
> Do you know the stories of Ada implementation in the early 80's?

Early Fortran implementations took a lot longer to write than they would
now because we (we software engineers that is) learned a lot about how to 
write compilers since then. The same can be said about the early Ada
implementations; many features in Ada 83 were not in use in other
mainstream languages, and so much was learned about compiling them. 
From my own experience as a user, I can assure you that C++ compilers of
the late 80's and early 90's were no great testament to the compilability
of that language.

While "argument from authority" is worthless, I have to say that I think
Robert Dewar is intimately familiar with several compiler implementations, 
and I would tend to believe his assessments of their relative complexities, 
since he has authored Ada, Cobol, and Spitbol compilers. 

Do you have a reasoned argument that an Ada compiler will be significantly 
more complex than a C++ compiler?

-- Brian