Version 1.2 of acs/ac-00110.txt

Unformatted version of acs/ac-00110.txt version 1.2
Other versions for file acs/ac-00110.txt

!standard 0.3(60)          05-02-09 AC95-00110/01
!class Amendment 05-02-09
!status received no action 05-02-09
!status received 04-12-14
!subject Revision timings
!summary
!appendix

From: Nick Roberts
Sent: Tuesday, December 14, 2004  2:10 PM

Pascal Leroy wrote:

> Nothing is easily added to the Standard at this point.  The core
> language is going into final review by the ARG on December 15th, and the
> only changes permitted at this point are either editorial or fixes to
> severe inconsistencies found during the review.  Don't even think of
> adding any new feature now.  (Of course, suggestions will be filed for
> consideration for the next revision, but that's at least 5 years from now.)

Of course (sorry).

I am quite excited by the prospect of a major new revision of the language
next year (or perhaps 2006?).

Would you be willing to give a quick rundown of the anticipated progress of
the revision (through WG9 and SC11?) from now on?  Fingers crossed that it
goes through smoothly!

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

From: Pascal Leroy
Sent: Wednesday, December 15, 2004  4:03 AM

The current plan is for the ARG to review the draft RM between now and the
end of April.  This RM (more precisely, the Amendment document that lists
only the changes) will then be sent to WG9 for consideration at its June
meeting in York.  If all goes well, the WG9 approval will be a mere
rubberstamping, and the Amendment will then be sent to SC22.  At this
point we expect things to go smoothly but slowly.  The difficult part will
be to go as fast as possible subject to the procedures and schedules of
SC22.  Jim Moore, the convener of WG9, is expert at dealing with the ISO
bureaucracy.  With some luck we could have an SC22 letter ballot next
Fall, with official publication at the beginning of 2006.

So chances are that from the ISO standpoint this will be Ada 2006.  But
from our perspective things will be pretty much cast in concrete by June.

Of course, that's if all goes according to plan.  If all doesn't go
according to plan, well, I suppose I'll lose interest...

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

From: Dan Eilers
Sent: Friday, December 17, 2004  4:23 PM

Will this official publication be the RM for Ada 2006, or will it
be just the list of changes since Ada 95 + TC1?

If only a list of changes is officially published, what will be the
proper way to refer to the language: Ada 06? or Ada 95 + TC1 + Amendment 1?

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

From: Jeff Cousins
Sent: Tuesday, December 21, 2004  10:50 AM

I haven't seen a reply to this, but I believe that a consolidated manual
will be published by Springer.  As to what its formal name will be, I'm
afraid that I don't know,  Springer published a consolidated manual for 2000
but it was still officially Ada 1995 + TC1 not Ada 2000.

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

From: Dan Eilers
Sent: Tuesday, December 21, 2004  12:21 PM

The situation with TC1 was quite different from the current situation,
so I hope it isn't used as a model.  TC1 was strictly a "bug-fix"
release of the language, adding essentially no new functionality.
Generally, bug-fix releases of products do not change the name of the
product.  Instead, we have for example Windows X with service pack Y,
or Ada95 with TC1.  This naming strategy conveys to the customers that
forward and backward compatibility issues are negligible, and that there
is a negligible need for retraining.

With the new Amendment, however, we have significant new functionality.
There will be a new Rationale document touting all the wonderful new
features.  Published Ada code will need to clarify whether or not it
requires an Ada 06 compiler.  Customers will need to specify when they
order compilers whether or not they require Ada 06.  So I think it's
important that the world knows that Ada has changed, and that this new
RM be officially called Ada 06 from ISO and everyone else's perspective.

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

From: Tucker Taft
Sent: Tuesday, December 21, 2004  1:08 PM

Because this is an "amendment" rather than a "revision" (for
a number of good reasons, the biggest being that a revision
would require a complete revamping of the format of the
manual), the official "ISO" document is a set of differences,
rather than a new manual.  On behalf of Ada-Europe,
Springer will be publishing the consolidated manual, which
I presume will have a title such as "Ada 2005" (or Ada 2006).

 From ISO's perspective, the name will be ISO/IEC-8652:2006,
I suspect, though I'm not sure.  Hopefully Randy could respond.
In general, ISO's nomenclature seems somewhat irrelevant to
what the community calls it.  From an ACATS testing and Ada marketing
point of view, the community will need to agree upon what
we call it, but it is probably too much to ask ISO to bless
whatever name we choose.

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

From: Randy Brukardt
Sent: Tuesday, December 21, 2004  1:32 PM

>  From ISO's perspective, the name will be ISO/IEC-8652:2006,
> I suspect, though I'm not sure.  Hopefully Randy could respond.

The Amendment document will likely be called ISO/IEC 8652:1995-AMD.1:2006
(or something like that). But ISO (*not* us) has the option of publishing an
updated standard or the various parts. Thus, exactly what they publish (and
therefore what it will be called) is pretty much out of our hands. We should
not be depending on that for the name of the language.

> In general, ISO's nomenclature seems somewhat irrelevant to
> what the community calls it.  From an ACATS testing and Ada marketing
> point of view, the community will need to agree upon what
> we call it, but it is probably too much to ask ISO to bless
> whatever name we choose.

Right. The current Fortran update is called Fortran 2003, even though
probably won't be published by ISO *this* year. If we agree to call this Ada
2005, then that is good enough for virtually all purposes.

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

From: Florian Weimer
Sent: Tuesday, December 21, 2004  3:12 PM

> Because this is an "amendment" rather than a "revision" (for
> a number of good reasons, the biggest being that a revision
> would require a complete revamping of the format of the
> manual), the official "ISO" document is a set of differences,
> rather than a new manual.  On behalf of Ada-Europe,
> Springer will be publishing the consolidated manual, which
> I presume will have a title such as "Ada 2005" (or Ada 2006).

Will this have any adverse affect on the freeness (free as in freedom)
of the electronic form?

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

From: Tucker Taft
Sent: Tuesday, December 21, 2004  10:08 PM

We have worked hard to ensure that the copyright will
allow free downloading, etc.  This is easier because
this is not an official ISO document.  The website:

    http://www.ada-auth.org

will probably continue to be a place where you can
get the latest, and final, documents.

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

From: Martin Krischik
Sent: Wednesday, December 22, 2004  3:04 AM

True! The eye catcheer "C99" was published 22.05.2000 and still everybody
calls is C99 and not C00 ;-) . Saying that - C2000 would have been a cool
name as well.

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

From: Dan Eilers
Sent: Wednesday, December 22, 2004  3:13 AM

> Because this is an "amendment" rather than a "revision" (for
> a number of good reasons, the biggest being that a revision
> would require a complete revamping of the format of the
> manual), the official "ISO" document is a set of differences,
> rather than a new manual.

I was under the impression that the consolidated RM that Randy
has been putting together is already in the format that he would
provide to ISO for publication if ISO so requests.  How does the
latest draft RM differ from the "completely revamped" format that
ISO actually requires?

> From ISO's perspective, the name will be ISO/IEC-8652:2006,
> I suspect, though I'm not sure.  Hopefully Randy could respond.

ISO procedures (15.5.1) are clear that when only an amendment is
published, the prior edition of the RM remains in print, so from
ISO's perspective the language would officially still be Ada95,
not ISO/IEC-8652:2006 as you suggest.

> In general, ISO's nomenclature seems somewhat irrelevant to
> what the community calls it.

Well, given that Ada is an ISO standard, it's not completely
irrelevant what ISO calls it.  The current version was called
variously Ada9X or Ada94 prior to the ink drying on the ISO
standard in January 1995, and since then it has always been
called Ada95 because that's what ISO calls it.

>                    From an ACATS testing and Ada marketing
> point of view, the community will need to agree upon what
> we call it, but it is probably too much to ask ISO to bless
> whatever name we choose.

Yes, the Ada community needs to agree on what we call it,
but that sort of agreement is normally accomplished by
using the name established by the owner of a standard.
We seem to be in uncharted territory wanting to name an
ISO standard something that we don't want to ask ISO to bless.

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

From: Maruis Amado Alves
Sent: Wednesday, December 22, 2004  5:31 AM

Please call it Ada 2005. This name is already in use in many items. It
would be difficult and in same cases impossible to change. And nothing
would be gained. And please keep the four digits. (To distinguish from
Ada 2105 :-)

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

From: Tucker Taft
Sent: Wednesday, December 22, 2004  6:30 AM

> I was under the impression that the consolidated RM that Randy
> has been putting together is already in the format that he would
> provide to ISO for publication if ISO so requests.  How does the
> latest draft RM differ from the "completely revamped" format that
> ISO actually requires?

No connection, unfortunately.  The completely revamped format,
if it were to follow ISO's current rules for standards, would
have all of the "interesting" material in chapter 3 of the
standard.  All the other chapters are essentially boilerplate,
references to other standards, etc.  By going the amendment route,
we avoid that massive restructuring.  ISO can still request
a consolidated document, but that should be produced by
a nearly automated process that merges the prior standard
and the amendment, and would not be expected by ISO to have the new
ISO structuring.

>>From ISO's perspective, the name will be ISO/IEC-8652:2006,
>>I suspect, though I'm not sure.  Hopefully Randy could respond.

As Randy indicated, the name is more likely to be something
like ISO/IEC-8652:1995-AMD.1 blah blah blah.

> ISO procedures (15.5.1) are clear that when only an amendment is
> published, the prior edition of the RM remains in print, so from
> ISO's perspective the language would officially still be Ada95,
> not ISO/IEC-8652:2006 as you suggest.

Right.  Randy hopefully clarified that.

>>In general, ISO's nomenclature seems somewhat irrelevant to
>>what the community calls it.
>
> Well, given that Ada is an ISO standard, it's not completely
> irrelevant what ISO calls it.  The current version was called
> variously Ada9X or Ada94 prior to the ink drying on the ISO
> standard in January 1995, and since then it has always been
> called Ada95 because that's what ISO calls it.

ISO doesn't call it Ada 95, but they did put the year 1995 on
their publication.  Other ISO standards such as C and Fortran
are known by names that reflect the year when the standard was
essentially complete, rather than the year the ISO process
finally completed.  So I think we have a choice.  ISO is
not about to legislate this.  The only danger would be
confusion, but we know ISO standards only come out about
every 5 years, so if we call it Ada 2005 while ISO ends
up with an official 2006 date on the Amendment document,
I don't think anyone will be confused.

>>                   From an ACATS testing and Ada marketing
>>point of view, the community will need to agree upon what
>>we call it, but it is probably too much to ask ISO to bless
>>whatever name we choose.
>
> Yes, the Ada community needs to agree on what we call it,
> but that sort of agreement is normally accomplished by
> using the name established by the owner of a standard.
> We seem to be in uncharted territory wanting to name an
> ISO standard something that we don't want to ask ISO to bless.

Not at all.  Fortran and C have charted this territory
quite completely, with little connection between the year
used in the common name and the year on the ISO document.

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

From: Jean-Pierre Rosen
Sent: Wednesday, December 22, 2004  7:57 AM

Note by the way that "Ada83" was the commonly used name for a standard
issued in 1987, as far as ISO is concerned...

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

From: Robert A. Duff
Sent: Wednesday, December 22, 2004  11:08 AM

The ink (toner?) actually dried in December of 1994.  That is, I printed
out the final version then.  (I had to get a print shop to specially cut
paper to A4 size; they had never heard of it, and I had to give them the
measurements!)  This version was delivered to ISO, and became official
in 1995.  I still think we should have called it "Ada 94" in normal
conversation.

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

From: Dan Eilers
Sent: Wednesday, December 22, 2004  12:17 PM

> No connection, unfortunately.  The completely revamped format,
> if it were to follow ISO's current rules for standards, would
> have all of the "interesting" material in chapter 3 of the
> standard.  All the other chapters are essentially boilerplate,
> references to other standards, etc.  By going the amendment route,
> we avoid that massive restructuring.

This "massive restructuring" might not turn out to be nearly as
much trouble as it sounds, as was learned with the similar
experience with the Ravenscar guide.  The 95 RM was already
restructured from 83 to use "sections" instead of "chapters",
so essentially the entire draft RM as it now stands could be
slotted directly into Chapter 3, with the addition of some
ISO boilerplate at the beginning.

This would seem to be a small price to pay in order to be able
to retire the Ada95 RM and legitimately call the new language
something other than Ada 95.

>                                        ISO can still request
> a consolidated document, but that should be produced by
> a nearly automated process that merges the prior standard
> and the amendment, and would not be expected by ISO to have the new
> ISO structuring.

Yes, and the ARG or WG9 or SC 22 or JTC 1 or any of the national
bodies can recommend that ISO do this.  Absent a recommendation
from the interested parties, ISO assumes that a SC that chooses
to go the amendment route does so precisely to avoid retiring
the previous RM.  So ISO is extremely unlikely to retire the 95 RM
without a recommendation to do so.

> ISO doesn't call it Ada 95, but they did put the year 1995 on
> their publication.

They put the year 1995 in the name of the standard: ISO/IEC-8652:1995,
not just somewhere on their publication.  A purchasing agent wishing
to procure the latest ISO Ada compiler who looks up the official name
from ISO's web site will find ISO/IEC-8652:1995, and when you see such
a RFQ, you won't know which language to bid.

> Not at all.  Fortran and C have charted this territory
> quite completely, with little connection between the year
> used in the common name and the year on the ISO document.

Yes, Fortran and C are famous for their off-by-one errors,
but that isn't the issue here.  We're talking about an off-by-
a-decade error.

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

From: Randy Brukardt
Sent: Wednesday, December 22, 2004  12:50 PM

> This "massive restructuring" might not turn out to be nearly as
> much trouble as it sounds, as was learned with the similar
> experience with the Ravenscar guide.  The 95 RM was already
> restructured from 83 to use "sections" instead of "chapters",
> so essentially the entire draft RM as it now stands could be
> slotted directly into Chapter 3, with the addition of some
> ISO boilerplate at the beginning.

The Ravenscar guide is a technical report; the rules are a lot more flexible
for them. None of the exceptions apply to a standard.

For standard, even the fonts are wrong in the current documents. For
instance, the body needs to be in Helvetica, and the examples in Times
Roman. We'd also have to create a separate definitions section and rewrite
every paragraph where there currently is a definition. The result would be
pretty much unreadable (look at 18009 for an example of what I mean); it
would no longer serve the dual purpose of a standard and a language
reference.

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

From: Dan Eilers
Sent: Wednesday, December 22, 2004  1:11 PM

Randy wrote:
> For standard, even the fonts are wrong in the current documents. For
> instance, the body needs to be in Helvetica, and the examples in Times
> Roman. We'd also have to create a separate definitions section and rewrite
> every paragraph where there currently is a definition. The result would be
> pretty much unreadable (look at 18009 for an example of what I mean); it
> would no longer serve the dual purpose of a standard and a language
> reference.

This misses my point.  Nobody cares whether the official ISO standard
is readable, since nobody buys that version.  The purpose served by
publishing a new official ISO standard is to retire the Ada95 RM
and Ada95 nomenclature.

If it is an excessive amount of effort to change the fonts and
rewrite the paragraphs with definitions, then it seems the ARG or
WG9 should start the ball rolling on a recommendation to ISO that
they publish the currently-formatted merged document.

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

From: Dan Eilers
Sent: Friday, January 7, 2005  6:28 PM

> True! The eye catcheer "C99" was published 22.05.2000 and still everybody
> calls is C99 and not C00 ;-) . Saying that - C2000 would have been a cool
> name as well.

According to:
  http://www.open-std.org/jtc1/sc22/wg14/www/standards

 "The current standard for Programming Language C is ISO/IEC 9899:1999,
   published 1999-12-01."

So I think everybody calls it C99 because ISO calls it ISO/IEC 9899:1999.

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



Questions? Ask the ACAA Technical Agent