Re: Perhaps there _is_ a conspiracy against Ada

[clines@delete_this.airmail.net (Kevin Cline)]

dewar@merv.cs.nyu.edu (Robert Dewar) wrote:

><<And just how did you attach callback functions to GUI objects
>since the language did not support passing functions or procedures as
>arguments?  Sure, each vendor defined a way to do that, but I had a
>requirement to port this program to multiple architectures and compilers.>>
>
>The ability to port a program does not mean that you require that the
>program be movable with zero changes, just that the parts that need
>changing are well defined and easily changed.

Well-defined?  Certainly.  Easily changed?  Not without resorting
to some macro preprocessor that would have confused Ada compilers and
debuggers ignorant of anything equivalent to # source line directives.

>In fact a number of compilers implemented the semi-standard for subprogram
>pointers defined by the CIFO specification, so programs written to the
>CIFO spec would in fact work on quite a variety of compilers.

Which ones, and when?  "A variety of compilers" didn't include the varieties
I was using.

>Note that if you required 100% portability with zero change, then you
>would say that no C program is portable, which is an exaggeration!

The techniques for writing portable C in 1991 were simple and well-known,
and a single set of source files could be made to work on a very large number
of hardware and operating system combinations.
Most of the portability problems for C programs were actually due to
variations in operating system services, which couldn't even be called in a
standard way using  Ada-83.  With no preprocessor, working around Ada-83's
missing features was considerably more aggravation.