Re: Ragged Array Proposal

[Robert Dewar <robert_dewar@my-deja.com>]

In article <37EBD2F3.E1E32672@ftw.rsc.raytheon.com>,
  Wes Groleau <wwgrol@ftw.rsc.raytheon.com> wrote:
> > In this particular case, there is a significant
implementation
> > burden (wait till you are really working on your Ada
compiler
> > to appreciate that :-). In particular, getting the debugger
to
> > understand this completely new type would be significant
work.
>
> In the cases of gdb and Apex Duo, where the same debugger
understands
> both C and Ada, then the debugger should be able to understand
>
>    char *fruits[] = { "apple", "orange", "pear" };
>
> So how much work would it be to handle an Ada syntax that has
the
> same underlying implementation?


Oops, I got mixed up, and thought that Nick had written this,
so please ignore my suggestion of implementing a debugger (that
was intended for Nick :-)

Wes, if you are seriously proposing that the ragged arrays
be implemented with a storage structure identical to that used
for fruits above, then you have suddenly enormously increased
the implementation burden.

That's because the C structure depends on the notion of simple
machine addresses that point to a null terminated string.

This means you have to introduce a whole new concept at both
the Ada semantic level (strings that are not allowed to contain
the null character), and at the implementation level (string
pointers with implicit bounds that are incompatible with normal
pointer-to-unconstrained).

It's doable, I suppose, but you have enormously increased the
complexity of the proposal, probably unintentially (that's the
trouble in language design, things that seem so simple and
straightforward turn out to be much more messy than you think).

Robert


Sent via Deja.com http://www.deja.com/
Before you buy.