Re: Seeking for papers about tagged types vs access to subprograms

["Randy Brukardt" <randy@rrsoftware.com>]

"Niklas Holsti" <niklas.holsti@tidorum.invalid> wrote in message 
news:av8pvtFehv2U1@mid.individual.net...
...
> It is interesting that Randy thinks his ideas regarding a future
> replacement for Ada resemble the typestate concept, but that the
> typestate concept as implemented in Plaid seems to be implementable in
> Ada 2012. Perhaps Randy's ideas go much further than this, however.

The main thing I was thinking about was some extension to the things that 
have to be compile-time analyzable (as Static Predicates are in Ada 2012). 
Probably the basis of them would remain subtypes and subprogram profiles 
(via preconditions and postconditions). For the Open example, that requires 
some way to encode the notion of "properties" in a statically understandable 
way. Perhaps you are right that discriminants would do the trick, but we'd 
want them to be "virtual" discriminants without any runtime cost.

I agree that you can get the effect of typestate analysis in Ada 2012 using 
discriminants, predicates, and pre/postconditions, but those would be 
checked at runtime. The key here for me is to require static detection of 
these errors, even when variables and unconstrained formal parameters are 
involved.

Anyway, I'm just musing here as opposed to having fully worked out ideas. So 
I could in fact be going down the wrong path.

                                                Randy.