>jim.fleming@bytes.com (Jim Fleming) wrote:
>> >Oh, that's an unfair challenge - first you'd have to find a compiler
>> >that supported "full" (draft) Ansi C++, and there aren't any ;-)
>> >
>> Hold on one second. The above can not be true. Fergus, are you trying to
>> say that there is no current implementation of the proposed ANSI C++
>> Standard.

 There are no compilers implementing the full ISO C++
language as described in the Committee Draft. There are language
features for which there are NO working implementations
available -- and perhaps not even any exprimental ones.

 That this is the case is a valid concern  -- that
it is the case however, is a consequence of the committee
being brave enough to engage in advanced language design
and _drive_ the quality of software upwards by international
consensus of representatives of vendors and users.

 I support this process and I think the Ansi/ISO
committee has done an excellent job here: the majority
of the extensions relate to templates and were championed by
Bjarne Stroustrup, for which he deserves considerable credit.

 I happen to think trying to freeze the C++ document
right now is a bit premature given the lack of solid implementation
experience. I do not think placing those features in the
language was incorrect.

 If I may explain -- templates as per the ARM we
a basic experimental feature. The committee agreed that
turing the experiment into reality was a good thing to do.
This involves a LOT more than just providing a precise
description of the features tentatively proposed in the
ARM -- it involved examining the evidence of existing
implementations and extending the semantics to reach
some kind of completion.

 To a very large extent the ability to correctly
support the kind of library STL is has driven this process.

 Summary: the idea ISO committees merely retroactively
standardise "existing practice" is completely out of synch
with modern needs. We MUST proactively create standards
for which no implementations exist because this is the
only way implementors will be prepared to commit the
resources to the implementation of the facilities.


--
        JOHN (MAX) SKALLER,         INTERNET:maxtal@suphys.physics.su.oz.au
 Maxtal Pty Ltd,
        81A Glebe Point Rd, GLEBE   Mem: SA IT/9/22,SC22/WG21
        NSW 2037, AUSTRALIA     Phone: 61-2-566-2189

In article <KANZE.95Jun23122606@slsvhdt.lts.sel.alcatel.de>,
James Kanze US/ESC 60/3/141 #40763 <kanze@lts.sel.alcatel.de> wrote:
>I don't think that the only question is whether the extension is
>implementable.  It would be nice if it had also been available in a
>wide circle so that it was tested with regards to usability.  It's
>only with experience that we will know whether the chosen syntax for
>namespaces, for example, was the right one.
[sig cut]

It's also a question whether a chosen _semantics_ for a feature is
correct.  The language specification, including the current draft,
does not specify the semantics of the language formally.  Without that,
compiler writers will invariable interpret the specification differently.


Adam Seligman
Formal Analysis of C++:
ftp://cs.williams.edu/pub/students/95als/main.ps or main.dvi

In article <DAM8qK.pC@kithrup.com>, mrs@kithrup.com says...
>
>In article <3s8afm$2cj@News1.mcs.com>,
>Jim Fleming <jim.fleming@bytes.com> wrote:
>>>Oh, that's an unfair challenge - first you'd have to find a compiler
>>>that supported "full" (draft) Ansi C++, and there aren't any ;-)
>>
>>Hold on one second. The above can not be true. Fergus, are you trying to
>>say that there is no current implementation of the proposed ANSI C++
>>Standard.
>
>I'll give you a hint, Fergus generally has a clue.
>
>If the above can not be true, name one compiler that supports
>templated members, namespaces, typename and all the other features
>list in the draft on my home page.  I dare you!  For an easier time,
>just name one shipping compiler that implements templated members.
>
>:-)
>
>(Yes folks, I love inciting Mr. Fleming, stand back and watch the fun!)
@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@

OK, so everyone agrees that there is no single compiler that implements
the entire proposed draft standard.

Evidently, people are comfortable with the fact that feature A was
tested on one compiler, feature B on another, and feature C on still
another. Also, people seem comfortable with the fact that most of the
vendors will have no problem pulling all of the features together into
their standard compliant implementation which some think will be ready
by 1996.

Does anyone have a feel for how much code (if any) will break between
now and then and also what developers should do in the interim. I am
most interested in the area of class libaries but I would assume that
this question will extend to language capabilities.

In retrospect, it appears that class libraries should have been part
of a separate standard but that is life and past history. Maybe someday
someone can explain how the proposed class library came to be. In the
meantime, it appears that some rapid convergence is required if people
are going to be able to write portable, reusable code and protect their
investments in tools and class libraries.

@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
--
Jim Fleming            /|\      Unir Corporation       Unir Technology, Inc.
jrf@tiger.bytes.com  /  | \     One Naperville Plaza   184 Shuman Blvd. #100
%Techno Cat I       /   |  \    Naperville, IL 60563   Naperville, IL 60563
East End, Tortola  |____|___\   1-708-505-5801         1-800-222-UNIR(8647)
British Virgin Islands__|______ 1-708-305-3277 (FAX)   1-708-305-0600
                 \__/-------\__/       http:199.3.34.13 telnet: port 5555
Smooth Sailing on Cruising C+@amarans  ftp: 199.3.34.12 <-----stargate----+
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\____to the end of the OuterNet_|

mrs@kithrup.com (Mike Stump) wrote:
>In article <3s8afm$2cj@News1.mcs.com>,

>If the above can not be true, name one compiler that supports
>templated members, namespaces, typename and all the other
features
>list in the draft on my home page.  I dare you!  For an easier
time,
>just name one shipping compiler that implements templated
members.
>

Template members for sure haven't been implemented in any
shipping compiler. And I guess this is one area where Fleming's
argument holds. I'm not sure if the implications of having
template members was fully understood specially with respect
to Inheritance and specialization and in the last meeting
compiler developers have mentioned problems that could occur.

>
>(Yes folks, I love inciting Mr. Fleming, stand back and watch
the fun!)

I guess Fleming now has enough ammunition to last a week.

- Kalyan

jim.fleming@bytes.com (Jim Fleming) wrote:
> >Oh, that's an unfair challenge - first you'd have to find a compiler
> >that supported "full" (draft) Ansi C++, and there aren't any ;-)
> >
> [snip]
> >--
> >Fergus Henderson
> >fjh@cs.mu.oz.au
> >http://www.cs.mu.oz.au/~fjh
> @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
>
> Hold on one second. The above can not be true. Fergus, are you trying to
> say that there is no current implementation of the proposed ANSI C++
> Standard.
>
> Are you saying that the language and associated class library have been
> proposed as a national and international standard and no one, not even
> the primary developers have *tested* the proposed standard...???
>
> I find this hard to imagine. C++ has been under development for 15 years.
> I believe that the ANSI standardization process was launched by Hewlett
> Packard about 5 years ago. It is hard to imagine that all of the features
> have not been tested 1,000 times and by this time most of the quirks have
> been found and fixed.
Yes. Companies write compilers. companies are driven by selling
products.  products have features users want/need.
The majority of users are not up in arms about having no GC on their
compilers.  The majority of (C++) users think portability is more
important than ease of use.  This is why the ANSI/ISO standardise,
this drives users to want portable features on their compilers, and
vendors put in the hours.

BS explains alot of this in the D&E of C++.

Dom

shepherd@debussy.sbi.com (Marc Shepherd) wrote:
>
> In article 12326@atlas.tntech.edu, rad6938@gemini.tntech.edu (Rad) writes:
> >
> >It seems obvious to me that what Fergus is trying to say is that no one
> >compiler implements the full (draft) ANSI-C++ standard.  This does not mean
> >that all the features weren't tested individually in various compilers.
> >(Standards usually involve a lot of combining features in several different
> >compilers.  In fact it would be stupid not to do this.)
> >
>
> While this is broadly correct, I *do* believe there are requirements in
> the draft ANSI/ISO standard that have not yet been implemented by anyone.
> The proposed locale/iostreams package is an obvious example.  Some of the
> latest namespace and template features are another.
>
> And yes, as a user who is committed to C++, this does make me a bit nervous.
Don't be: members like BS make sure there submissions *are*
implementable (and effecient too)..:)

Dom

In article <3s8afm$2cj@News1.mcs.com>,
Jim Fleming <jim.fleming@bytes.com> wrote:
>>Oh, that's an unfair challenge - first you'd have to find a compiler
>>that supported "full" (draft) Ansi C++, and there aren't any ;-)
>
>Hold on one second. The above can not be true. Fergus, are you trying to
>say that there is no current implementation of the proposed ANSI C++
>Standard.

I'll give you a hint, Fergus generally has a clue.

If the above can not be true, name one compiler that supports
templated members, namespaces, typename and all the other features
list in the draft on my home page.  I dare you!  For an easier time,
just name one shipping compiler that implements templated members.

:-)

(Yes folks, I love inciting Mr. Fleming, stand back and watch the fun!)

In article <3sctc7$8tp@snlsu1> Dom De Vitto
<devitto@ferndown.ate.slb.com> writes:

|> shepherd@debussy.sbi.com (Marc Shepherd) wrote:

|> > While this is broadly correct, I *do* believe there are requirements in
|> > the draft ANSI/ISO standard that have not yet been implemented by anyone.
|> > The proposed locale/iostreams package is an obvious example.  Some of the
|> > latest namespace and template features are another.

|> > And yes, as a user who is committed to C++, this does make me a bit nervous.
|> Don't be: members like BS make sure there submissions *are*
|> implementable (and effecient too)..:)

I don't think that the only question is whether the extension is
implementable.  It would be nice if it had also been available in a
wide circle so that it was tested with regards to usability.  It's
only with experience that we will know whether the chosen syntax for
namespaces, for example, was the right one.
--
James Kanze         Tel.: (+33) 88 14 49 00        email: kanze@gabi-soft.fr
GABI Software, Sarl., 8 rue des Francs-Bourgeois, F-67000 Strasbourg, France
Conseils en informatique industrielle --
                              -- Beratung in industrieller Datenverarbeitung

In article <9517210.11940@mulga.cs.mu.OZ.AU>, fjh@munta.cs.mu.OZ.AU says...
>
>chase@centerline.com (David Chase) writes:
>
>>In article <3r0hd8$r35@nova.umuc.edu>, COATES@EUROPA.UMUC.EDU (Ell)
writes:
>>
>>|> There are 3rd party tools for almost any feature C++ doesn't have
>>|> natively.
>>
>>Name one supported 3rd party tool that supports garbage collection for
>>"full" Ansi C++ ("full" meaning conforming to the draft standard.)
>
>Oh, that's an unfair challenge - first you'd have to find a compiler
>that supported "full" (draft) Ansi C++, and there aren't any ;-)
>
[snip]
>--
>Fergus Henderson
>fjh@cs.mu.oz.au
>http://www.cs.mu.oz.au/~fjh
@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@

Hold on one second. The above can not be true. Fergus, are you trying to
say that there is no current implementation of the proposed ANSI C++
Standard.

Are you saying that the language and associated class library have been
proposed as a national and international standard and no one, not even
the primary developers have *tested* the proposed standard...???

I find this hard to imagine. C++ has been under development for 15 years.
I believe that the ANSI standardization process was launched by Hewlett
Packard about 5 years ago. It is hard to imagine that all of the features
have not been tested 1,000 times and by this time most of the quirks have
been found and fixed.

@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
--
Jim Fleming            /|\      Unir Corporation       Unir Technology, Inc.
jrf@tiger.bytes.com  /  | \     One Naperville Plaza   184 Shuman Blvd. #100
%Techno Cat I       /   |  \    Naperville, IL 60563   Naperville, IL 60563
East End, Tortola  |____|___\   1-708-505-5801         1-800-222-UNIR(8647)
British Virgin Islands__|______ 1-708-305-3277 (FAX)   1-708-305-0600
                 \__/-------\__/       http:199.3.34.13 telnet: port 5555
Smooth Sailing on Cruising C+@amarans  ftp: 199.3.34.12 <-----stargate----+
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\____to the end of the OuterNet_|

In article <3s8afm$2cj@News1.mcs.com>, jim.fleming@bytes.com (Jim Fleming) writes:
>In article <9517210.11940@mulga.cs.mu.OZ.AU>, fjh@munta.cs.mu.OZ.AU says...
>>
>>chase@centerline.com (David Chase) writes:
>>
>>>In article <3r0hd8$r35@nova.umuc.edu>, COATES@EUROPA.UMUC.EDU (Ell)
>writes:
>>>
>>>|> There are 3rd party tools for almost any feature C++ doesn't have
>>>|> natively.
>>>
>>>Name one supported 3rd party tool that supports garbage collection for
>>>"full" Ansi C++ ("full" meaning conforming to the draft standard.)
>>
>>Oh, that's an unfair challenge - first you'd have to find a compiler
>>that supported "full" (draft) Ansi C++, and there aren't any ;-)
>>
>[snip]
>
>Hold on one second. The above can not be true. Fergus, are you trying to
>say that there is no current implementation of the proposed ANSI C++
>Standard.
>
>Are you saying that the language and associated class library have been
>proposed as a national and international standard and no one, not even
>the primary developers have *tested* the proposed standard...???
>
>I find this hard to imagine. C++ has been under development for 15 years.
>I believe that the ANSI standardization process was launched by Hewlett
>Packard about 5 years ago. It is hard to imagine that all of the features
>have not been tested 1,000 times and by this time most of the quirks have
>been found and fixed.

It seems obvious to me that what Fergus is trying to say is that no one
compiler implements the full (draft) ANSI-C++ standard.  This does not mean
that all the features weren't tested individually in various compilers.
(Standards usually involve a lot of combining features in several different
compilers.  In fact it would be stupid not to do this.)

----------------------------------------------------------------------------
 Richard Deken                   Graduate student in electrical engineering
 PGP public key available      Tennessee Technological University
 Internet: rad6938@gemini.tntech.edu        Cookeville, TN, USA

In article 12326@atlas.tntech.edu, rad6938@gemini.tntech.edu (Rad) writes:
>
>It seems obvious to me that what Fergus is trying to say is that no one
>compiler implements the full (draft) ANSI-C++ standard.  This does not mean
>that all the features weren't tested individually in various compilers.
>(Standards usually involve a lot of combining features in several different
>compilers.  In fact it would be stupid not to do this.)
>

While this is broadly correct, I *do* believe there are requirements in
the draft ANSI/ISO standard that have not yet been implemented by anyone.
The proposed locale/iostreams package is an obvious example.  Some of the
latest namespace and template features are another.

And yes, as a user who is committed to C++, this does make me a bit nervous.

---
Marc Shepherd
Salomon Brothers Inc
shepherd@schubert.sbi.com The opinions I express are no one's but mine!