| >1) Has anyone got any rough figures on how many man-months or whatever
>would be involved in converting about 33,000 lines of CORAL-66 code
>to Ada ASSUMING no extra functionality is required ?
Depends very much on the sytyle of Coral-66 programming. Coral-66
deliberately encourages the programmer to use pointer arithmetic all
over the place. It can be even worse than C, as the language
definition includes details on the relationship between the addresses
of different stack-based variables within a given stack frame. If the
code you're looking to port makes use of features like this, it's going
to be a tough job.
We used to market a Coral-66 compiler for VMS. That'd probably be a
better solution that trying to re-write a large Coral-66 system in any
other language.
John
|
| There are other CORAL 66 compilers for VMS - Systems Designers
used to have one, and even now as EDS-SCICON they would be worth
asking.
Generally, to translate from a less rich language like CORAL 66
or FORTRAN to Ada is a bad idea in principle, because you are
bound to get "Adatran" - no translator is going to be smart enough
to create Ada packages, tasks, etc. out of thin air. So you will
end up using the "CORAL 66 subset" of Ada, and you wouldn't be
a whole lot better off.
The "right" thing to do is to reimplement using Ada as a design
language. Then the application will gain in maintainability.
/Tom
|