!standard  3.5(35)                              10-02-19    AC95-00191/01
!class Amendment 10-02-19
!status received no action 10-02-19
!status received 09-11-18
!subject  Image attribute for objects
!summary

!appendix

From: Robert Dewar
Date: Wednesday, November 18, 2009  10:39 AM

Since we seem to have so much energy when it comes to discussing simple things,
how about the following:

     X'Image

allowed to be applied to objects as well as types (like the 'Img attribute of
GNAT).

I would say, don't reply to this unless you agree, that way if no one agrees,
nothing happens

****************************************************************

From: Randy Brukardt
Date: Wednesday, November 18, 2009  1:12 PM

Procedurally, the deadline for starting new proposals of any size by e-mail was
at noon (Wisconsin time) on November 3rd and in person at the end of the meeting
on November 8th (2:30 p.m. Tampa time). This is small enough that I'd be
inclined to ask the full ARG vote to waive that requirement, but that would have
to be done before we could consider it formally.

I'd be somewhat interested in the idea, my only concern would be the additional
complications in resolving the prefix of the attribute. I can't think of a
problem with that, however.

****************************************************************

From: Gary Dismukes
Date: Wednesday, November 18, 2009  2:25 PM

> I would say, don't reply to this unless you agree, that way if no one
> agrees, nothing happens

I would be in favor of it.

(And vote for waiving the rule of "no more feature suggestions" for this case:-)

****************************************************************

From: Tucker Taft
Date: Wednesday, November 18, 2009  2:34 PM

For what it's worth, AdaMagic's front end supports the 'Img attribute in the
same way as GNAT.  By implication, Green Hills also supports it (don't know
about Aonix, as they went off maintenance years ago).

Dare I ask about 'Wide_Image and 'Wide_Wide_Image? ;-)

I agree this is OK to consider, particularly since its equivalent is already
supported and widely used with at least two compilers, and perhaps more.  It
would be nice to remove this source of non-portability.

****************************************************************

From: Bob Duff
Date: Wednesday, November 18, 2009  2:46 PM

I vote the same way as Gary.

If it's already implemented in compilers, then there's no "implementation cost"
concern.

****************************************************************

From: Robert Dewar
Date: Wednesday, November 18, 2009  2:55 PM

The only reason GNAT called it Img was concerns that extending a standard
attribute was non-conforming.

****************************************************************

From: Robert Dewar
Date: Wednesday, November 18, 2009  2:57 PM

> Dare I ask about 'Wide_Image and 'Wide_Wide_Image? ;-)

I would include Wide_Image and Wide_Wide_Image if it becomes official. We never
did Wide_Img, since we never had a requirement for such :-)

****************************************************************

From: John Barnes
Date: Wednesday, November 18, 2009  4:48 PM

Now that is a good suggestion and indeed one of the ones I put in my shopping
list.

Let's do that and ease off on the assignment stuff which I find rather
disturbing. And my mailbox floweth over!!

****************************************************************