Re: Binding a type to a union.

[kilgallen@eisner.decus.org (Larry Kilgallen)]

In article <1103_943541468@DZOG-CHEN>, falis@ma.aonix.com (Ed Falis) writes:
> On Thu, 25 Nov 1999 14:12:01 GMT, kilgallen@eisner.decus.org (Larry Kilgallen) wrote:
> 
>> > 
>> > No one said this was a standard pragma, it is not! All I said
>> > was that it was not GNAT specific. Look it up in the Aonix
>> > documentation, or the GNAT documentation, or the Greenhills
>> > documentation etc. The GNAT implementation is slightly more
>> > restrictive than the Intermetrics one I believe.
>> 
>> It must be a _lot_ more _documented_, as I find no reference in the
>> Aonix Annex M section on Implementation-define pragmas (page 1-3
>> of UD/UG/A0000-05625/001, dated Mar97, the most recent version
>> distributed by Aonix to subscription customers).
>> 
>> Larry Kilgallen
> 
> Thanks for pointing out the error in Annex M - I'll file a documentation problem
> report.
> 
> Unchecked_Union and C_Pass_by_Copy are documented in the "Tour of 
> ObjectAda" document in the section on interfacing to other languages.

Indeed, now that you point them out, I find them in that document.
It does require turning to the right section, as that document
has no index at all.  But what would really be required for an
efficient lookup would be a master index covering all 12 of the
ObjectAda documents, and even my favorite documentation tool does
not handle the Master Index problem adequately.

=====

It might be good if Ada implementors when documenting pragmas
that are implementation-defined could indicate that they know
of other implementations of a pragma by the same name.

Given Robert's comment about variations in the degree of
restriction on this pragma between implementations, what
is the general process by which people see a useful but
not standard pragma evolving into a standard one (not
this paragma in particular).  Will we ultimately likely
see the more-restrictive or the less-restrictive variant
adopted ?

Larry Kilgallen