(j3.2006) a question on allowing internal procedures as procedure targets

[Malcolm Cohen]

Michael Ingrassia wrote:
> Oops, did I say 12.5.2.3?  I meant 12.6.2.4.
>
> Actually, 12.6.2.4 may not be saying quite what it ought (or I'm just not
> reading it correctly).  The last sentence of p4 says
>   
Please read what host instance is.  It's the instance of the host bound 
to the procedure pointer (or dummy).
>
> which seems to say that f gets access to the entities to the host instance
> of func !!!  Cool!  f is not limited to accessing entities from its host,
> but can access entities from ANY scoping unit which can declare a
> procedure pointer and call externFn !
>   
Obviously it does not seem to say that if you use the actual meaning of 
"host instance" rather than making one up.
> That's gotta be wrong I think.
>   
"host instance" does appear in the index...

...I could make it a defined term, but the definition would not be 
easier to understand than the normative text that says how it comes 
about and what it's used for...

...and I would have thought that having a strange term that's being used 
as the "static link" consistently would make a reader think "hmm, maybe 
this term means static link and not something else I just made up - 
perhaps I should look it up".  [P.S. I don't think "static link" is a 
better term than "host instance", either.]

BTW, there is a giant clue that "host instance" does not mean the 
procedure in which the procedure pointer is declared, because a 
procedure pointer *is not a scope* and *does not have a host*.

Cheers,
-- 
......................Malcolm Cohen.