Re: Subprogram types vs. "limited access" (was Re: Do we need "Mission-Critical" software? Was: What to Do?

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

In article <93npfn$13d$1@nnrp1.deja.com>,
  mark_lundquist@my-deja.com wrote:
> In article <93mqhh$4gl$1@nnrp1.deja.com>,
>   dmitry6243@my-deja.com wrote:
> > In article <93l6ut$pvf$1@nnrp1.deja.com>,
> >   mark_lundquist@my-deja.com wrote:
>
> > > Java wanted to get rid of pointer-related bugs.
Unfortunately, you
> > > kind of have to have pointers!
> >
> > Yes, but rare.
>
> I meant for linked data structures... you need by-reference
> semantics, or in a by-value language you need pointers.

No, that's not correct. Look at SETL, this is an example of
a language with absolutely 100% pure value semantics, and it
is perfectly easy to represent "linked data structures".
Consider for example a collection of nodes organized into
a binary tree.

SETL would represent this as a triple of values

the set of nodes
a map from parents to left-sons
a map from parents to right-sons

Each of these is treated will value semantics.

SETL itself is an imperative language, so you can have these
three values in variables, and modify them, but you can apply
this same viewpoint to a completely applicative language,
MIRANDA is an example.

I think what you say is true in the realm of current languages,
which are all at a very similar rather low semantic level, but
as a general statement, it is too sweeping.

(I am NOT suggesting removing pointers from Ada :-)


Sent via Deja.com
http://www.deja.com/