Re: Automated conversion to C++?

[Jacob Sparre Andersen <sparre@nbi.dk>]

Dr. Adrian Wrigley wrote:
> On Fri, 15 Feb 2008 14:33:08 +0100, Jacob Sparre Andersen wrote:

> If you have to supply a C++ program semanticly equivalent to the
> binary you deliver, the best choice is to write it in pure C (or
> C++).

The requirement for C++ has not been formulated in such strict terms.

If I come to the conclusion that writing directly in one of the
C-languages is the only option, I'll just skip the project.

> You are guaranteed equivalence.  You need find no translators.  You
> have no mismatches in runtime. The risk is low.

But my working environment will be unbearable.

> C++ makes so many Ada capabilities a real nuisance to implement.
> Basics like enumerations with constraint checks, fixed point types,
> floating point attributes, integer over/underflow exceptions need
> loads of messy C++ (often with specialized headers) to implement.

I don't care about how messy the C++ code is going to be.  We'll most
likely be running the Ada version internally.  The C++ code is only a
matter of public perception.  The project lead is of the opinion that
publishing the system with C++ code will make more scientists
interested in the system.

> If a customer is paying to get C++ source, don't write it in Ada -
> they presumably don't care for a robust, low-defect solution.  If
> they did, they would put constraints on the C++ (eg must run at
> least 80% of the speed.  Use no more than 1.3x the memory)

It is not a matter of code quality.  It is a matter of a decent work
environment for myself.

Thanks for the advice.

Jacob
-- 
�It will not be forever. - It will just seem like it.� -- Death