Re: named control statements (was: Future with Ada)

[Jeffrey Carter <jrcarter@acm.org>]

Wes Groleau wrote:
> 
> > > Would it be attractive to be able to do something like:
> > >
> > > Outer_Check: if (Some_Boolean) then
> > >     Inner_Check: if (Some_Other_Boolean) then
> > >         Some_Statement ;
> > >     end if Inner_Check ;
> > > end if Outer_Check ;
> >
> > Hmmm, looks exactly like Fortran 95...(except ; is optional)
> 
> That's OK.  Nothing wrong with borrowing a good idea
> from another language.  Others have been borrowing good
> ideas from Ada for almost twenty years.....

Something like this was part of the Red language, IIRC. There was also
something like this in the early forms of 9X, IIRC; not surprising,
since 9X was done by Intermetrics, who developed Red. I guess it was one
of many things considered not important enough to make it into Ada, like
"return when".

At the Ada Launch (1980 Dec 10), Ichbiah, Barnes, and Firth talked about
the "comb structures" in Ada. These are things like

procedure X is
|
begin
|
exception
|
end X;

package X is
|
private
|
end X;

and also things like

if X then
|
elsif Y then
|
elsif Z then
|
else
|
end if;

case X is
when ...
|
when ...
|
when ...
|
end case;

These look like combs: they have a backbone on the left and teeth
projecting to the right. If we repeat names on the inner parts of the
combs for subprograms, tasks, and packages (begin, exception, and
private), then it seems we should like to put them on the inner parts of
conditional statements (elsif, else, and when), but it's not clear where
they should go. Would we want

elsif Z then (X)
when A => (X)

or

X : elsif Z then
X : when A =>

? Maybe it would be best to be inconsistent and not allow them at all.
But I would certainly like compiler-checked names on as many parts of as
many structures as possible.

-- 
Jeff Carter
"Hello! Smelly English K...niggets."
Monty Python & the Holy Grail