ANSI/ISO C++ compliance of VC++ 7.0 sucks! ...

MS could not care less than standard compliance. Standards are things
which enable different products to interoperate. MS wants you to buy a
copy of Visual Studio for each developer and stay bound to them FOREVER.
Expect lots of proprietary extensions in VC7 and not standards compliance.

> Shiv: Anyone who really knows is under NDA and cannot discuss it.

Um, Microsoft has discussed standards compliance on the DOTNET mailing list
and they said that VC++.NET will NOT be completely standards compliant but
will be more standards compliant than previous versions.

Robert Scoble
http://www.vslive.com <<--The Fall's Largest Visual Studio Conference

###

> Yeah but exactly what will they support? And where is this
> .NET mailing list? How does one subscribe to it?

They didn't say.

You must visit my links list. I have more than 140 links to various ".NET"
resources at http://www.dev-archive.com/free/press/2000/vsnet/vsresources.asp

In particular, visit the Mailing List page at
http://www.dev-archive.com/free/press/2000/vsnet/vsresources-2.asp

The Developmentor DOTNET mailing list is right at top.

The most important questions and answers are at
http://www.dev-archive.com/free/press/2000/vs-qalist.asp

Robert Scoble
http://www.vslist.com <<-- The Fall's Largest Visual Studio Conference

###

"Shiv Shankar Ramakrishnan" <rshivs@hotmail.com> wrote in message
news:398e871d@news.dev-archive.com...
> How can VC++ users take all this lying down? If we don't ask
> for better std compliance and if we don't complain loudly then
> they are not going to give it to us. Or is it that the average
> VC++ user is pathetially unaware of these things that he
> doesn't care? That would be a very sad and sorry state of
> affairs. The day you try to port your C++ code to other
> compilers and/or platforms you will have major headaches.

I'm going to have to vote for pathetically unaware. I'd guess that most C++
programmers are under the same "grind-it-out-quick-and-dirty" pressure that
most every other programmer's under. No matter how much you care about
doing it right, you just don't have the time to "move-forward"(as I hear
they say in upper management) and learn about all those really great
language features that could give you long-term benefit. Because that's the
case(or they really just don't care), the standards compliance people are
probably pretty small voices in the crowd.

Sorry the previous posting got posted before I could finish.
Please read this one for the complete post.

> Are you sure you're using an updated compiler (as opposed

Yes I am dead sure. Yes I do know that once the VC++7 compiler
is installed the VC6 IDE still runs the older compiler and that
to run the new one you have to do it from the command line
provided you have registered the environment variables that
came with the NGWS SDK.

> Open a command prompt window, navigate to Bin directory under Visual
> Studio/VC and type "cl" (if you get an error, run vcvars32.bat first).
That
> should tell you the version of the actual compiler you are using. For my
> copy of vs6 (sp3) it's "12.00.8168"

Just to satisfy you the new VC++ 7 (when I run from cmd line)
gives me a version of - "13.00.8905" Let me also tell you where
that runs from -
E:\WINNT\ComPlus\v2000.14.1812\
that is
%SystemRoot%\ComPlus\v2000.14.1812

And thanks a lot for explaining in excruciating details on
how to do all this! What do you take me for? An average VC++
user who is absolutely clueless about all these things? But
then I doubt that even that possible conjecture of your would
be true or palatable to ppl :)
Thanks,
Shiv

Have you tried testing if VC7 now supports covariant return types?

--
Antony Andrews

Shiv Shankar Ramakrishnan wrote in message <398c6142@news.dev-archive.com>...
>Hi All,
>I've been playing with the Visual C++ 7 compiler which is part
>of the .NET SDK from the MSDN subscriber download. I was very
>very unhappy to see that they haven't progressed a lot in terms
>of ANSI/ISO C++ Std compliance.
>For example there isn't still full support for the following -
> * Member templates (With defns outside class body)
> * Partial template ordering
> * Partial template specialization
>For heavens sake all other compilers like Borland C++ 5.5 and
>GCC 2.95.2 (and for that matter MS's arch enemy Sun's Forte C++
>6.0 also) seem to support all these features. VC++ is a real
>good compiler but it will look pretty lame when ppl point this
>out. I really like C# also and they have done a great work
>there to save us from the farce that Java is. But still how
>could they neglect C++ like this?
>Don't tell me those guys don't have the ability to do it. Just
>think in what light people are going to paint Microsoft's
>recent Standards embracing postures when they talk about this
>one. Does anyone know if they are going to support these
>features by release time? All these are very important for
>things like STL and all. If these features aren't really
>important then I am going to dump Visual C++ for C++
>development at least. I've had great hopes from VC++ 7 but they
>seem to be dashing it all.
>Thanks,
>Shiv
>
>
>

I am the person that you quote (very indirectly, since you are quoting a web
site that has text that was culled from discussions on the .Net mailing list).

I don't really understand why you would rather have no answer than the honest
answer I tried to give in regards where we are with C++ Standards compliance.

I am one of the Microsoft representatives to J16 and I am reasonably aware
of C++ standards issues, including what parts of STL require partial specialization
for full conformance. I said that we would be releasing a very conformant
STL implementation, and I stand by that. Like always, if you have constructive
feedback on what user scenarios are impacted by the iterator traits issue,
we would love to hear them. We base almost all of our product feature decisions
on exactly such feedback.

We did not have the resources for VC.Net 7.0 to get to the few major areas
in template support where Visual C++ has no real support. The 3 you mention
below are in that category. Since you are posting in a newsgroup about .Net,
you should probably have a good idea on where we did spend a majority of
our resources in this release.

All the members of the C++ team consider standards conformance a very important
issue, and we don't like the fact that we couldn't do more in this release.
Planning is getting underway for the next version of Visual C++ and adding
the missing pieces is a top issue on the agenda.

I should really stress that feedback DOES matter. Feedback however is much
more effective when given directly to Microsoft Product Support or one of
our feedback aliases or on one of the Microsoft monitored newsgroups. Someone
just happened to have forwarded me your message. It wasn't immediately obvious
to me that reading a newsgroup called C#.general on a largely VB oriented
site would be a good way to find good customer feedback on Visual C++ Standards
compliance.

If you or any of the other forum participants want to address concerns with
standards compliance in Visual C++, feel free to send me an email.

Ronald Laeremans
Program Manager
Microsoft Visual C++

"Shiv Shankar Ramakrishnan" <rshivs@hotmail.com> wrote:
>> They didn't say.
>
>Apparently they did say a wee bit. But I wish they hadn't said
>anything and put their foot in their mouth.
>
>> The most important questions and answers are at
>> http://www.dev-archive.com/free/press/2000/vs-qalist.asp
>
>Ok lemme explain what I meant. Go to the above page and read
>these two questions there -
>
>Do you know which STL/version is going to be shipped with VC7?
>Does VC.NET support ANSI/ISO C++? Can VC.NET use Blitz++?
>
>Ronald Laeremans of Microsoft doesn't sound convincing at all
>if you analyze what he says. As answer to the first question
>he says that we will ship a very conformant version of STL. And
>in the next one he says that they won't be able to give partial
>template specialization in VC7. Now probably what he doesn't
>know is that it is *not* possible to have a conformant STL
>without that. You see the standard says that you need to give
>out iterator traits for all iterators for a number of functions
>to work properly in <algorithm>. Now you *cannot* do that fully
>without partial template spcialization. That happens coz the
>only way to give iterator traits for ordinary pointers is by
>partial template specialization.
>But most poor clueless VC users don't know anything about that
>and are being taken for a ride when they say that STL will
>be conformant. Why promise something you cannot and more impor-
>tantly *will not* deliver? Most advanced users of C++ have been
>crying for better template support in VC for a long time now.
>Doesn't look like coming. Its all so sickening!
>Thanks,
>Shiv

Hi Ronald,

> I am the person that you quote (very indirectly, since you are quoting a
web
> site that has text that was culled from discussions on the .Net mailing
list).

First of all I am very impressed to see you reply to my post
here. For all the time that I have read the VC++ STL newsgroup
I never saw any microsoft person reply there. Secondly my
sincere apologies for making such abrasive posts :) You see
I had more than one reason. Firstly I *did* send feedback.
Please check up the mails for ngwssdk@microsoft.com You will
find a mail from me there where I start by appreciating the
good work you did with template error messages etc and then I
talked about std compliance issues. And that was the email
address given in the SDK for feedback. I waited for a few days
before making this post of mine since I did not even get an
ack for my message. Also I have long asked along with others
on the STL list about std compliance in the next VC. *No one*
ever gave any hints there. And that forum cannot be faulted
since most of the template related issues are discussed there.
You also I am very sure know about the mess that happened due
to the dinkumware law suit due to which VC++ STL users had to
suffer for so long. If you again check up on VC++ support you
will find a number of mails from me there asking why the latest
updated dinkumware Std C++ lib was not being shipped with the
VC SP's. I never got any convincing answers. In fact I only
got any answer from VC++ support after I raised a stinker about
it in the STL newsgroup and other ppl also asked. So after
all this I was really frustrated as to why are things like this
and this was the reason I made such a post. I had to get my
message through some way and none of it seemed to work. And
you know something? When I had complained to VC++ support at
that time ppl outside the US weren't even allowed to access the
web based support! I was told to give feedback through the
Indian MS ppl. Which also I did. But nothing came. That is when
I went to the STL list and guess what after the whole issue at
that time as part of the reply they opened up the web based
stuff for all. This was some time last year BTW. So after I got
no reply for my mail about the feedback on VC++ 7 I chose the
only option left out. I do very much agree that my message used
harsh and abrasive language and I do really apologize but based
on what I had been through earlier I was very frustarted.

But now I have a lot of respect for you that you chose to
answer me so wisely and aptly and I shall now calmly answer all
that I can to your message.

> I don't really understand why you would rather have no answer than the
honest
> answer I tried to give in regards where we are with C++ Standards
compliance.

Yes I know you did give an honest answer. But rather that I did
not like talk about a very conformant STL without clearly
spelling out what exactly that meant.

> I am one of the Microsoft representatives to J16 and I am reasonably aware
> of C++ standards issues, including what parts of STL require partial
specialization
> for full conformance. I said that we would be releasing a very conformant
> STL implementation, and I stand by that. Like always, if you have
constructive

I confess there was room to give a lot of leeway in interpreti-
ng what very conformant means, but I chose to play on that
deliberately to provoke some reaction :) Sorry. It wasn't meant
as an affront to your knowledge or anything at all. It just
happened that you were the one talking about C++ there and it
wasn't meant to be explicitly pointing to you in any way.

> feedback on what user scenarios are impacted by the iterator traits issue,
> we would love to hear them. We base almost all of our product feature
decisions
> on exactly such feedback.

But the problem is that where and how do you give it?! A number
of STL users tried to talk a lot about Std C++ issues on
the STL newsgroup but the only person to ever say something
there has been P.J. Plauger himself and we never saw any MS
person say anything there. Infact Plauger himself said a number
of time that he knew the issues and that a number of his friends
knew the issues in the VC team but that he couldn't talk about
it and all. Why the great secrecy? You could at least as users
let us know what we could expect in the next VC when we all
asked. But no, we had to find it the hard way by using the
pre-beta compiler. And then when I don't find things about it
and send a mail asking about them (albiet a bit gruff mail :)
I get no reply at all. And then based on past experience, I just
got ... And due to all this I had long gaven up talking about
things on the STL list since it never helped.

> We did not have the resources for VC.Net 7.0 to get to the few major areas
> in template support where Visual C++ has no real support. The 3 you
mention
> below are in that category. Since you are posting in a newsgroup about
..Net,
> you should probably have a good idea on where we did spend a majority of
> our resources in this release.

Yes yes I know you have all worked long and hard on the .Net
features and I congratulate you all on coming up with a very
nice concept and yes I know it takes a lot of work. But still
what I did not like is that you never talked about Std
compliance issues a lot and that left us wondering if you ppl
were serious about it.

>
> All the members of the C++ team consider standards conformance a very
important
> issue, and we don't like the fact that we couldn't do more in this
release.
> Planning is getting underway for the next version of Visual C++ and adding
> the missing pieces is a top issue on the agenda.

Thanks. I am pretty releived about that. It would have been
nice if you could have reassured us some time before.

> I should really stress that feedback DOES matter. Feedback however is much
> more effective when given directly to Microsoft Product Support or one of
> our feedback aliases or on one of the Microsoft monitored newsgroups.
Someone

You probably know by now that I believe in that as much as you
do or why would I be complaining so much :)
Now honestly last year I had sent mails to MS product support
and that didn't seem to work :(

> just happened to have forwarded me your message. It wasn't immediately
obvious
> to me that reading a newsgroup called C#.general on a largely VB oriented
> site would be a good way to find good customer feedback on Visual C++
Standards
> compliance.

Very valid point. Only that in my experience it never seemed to
help on the STL newsgroup. Now I do not know if that is moni-
tered by MS or not but it certainly is on msnews.microsoft.com
That is why I deliberately chose to seek out where I would be
heard. And someone told me that these .Net newsgroups etc where
where a number of MS people were hanging out. So I entirely
skipped the option of posting on microsoft.public.vc.language
or microsoft.public.vc.stl. Okay maybe I might have got some
response had I posted there but based on past experience I
thought that might be a waste of time. So I knew that the
C++ ppl would be there for the C# stuff also and that why I
posted on C# newsgroup. Maybe I am guilty here as charged.

> If you or any of the other forum participants want to address concerns
with
> standards compliance in Visual C++, feel free to send me an email.

Yes very definitely. And thanks a lot for opening a channel
of communication. Once again I am sorry for being so sensatio-
nalist in my post :) But I was looking for some way of getting
a response. And my points were valid even though the way I put
it across wouldn't be considered well mannered I guess :)
Nothing against you per se. I think you did a good job in your
reply to me here.
Thanks,
Shiv

>
> Ronald Laeremans
> Program Manager
> Microsoft Visual C++
>
> "Shiv Shankar Ramakrishnan" <rshivs@hotmail.com> wrote:
> >> They didn't say.
> >
> >Apparently they did say a wee bit. But I wish they hadn't said
> >anything and put their foot in their mouth.
> >
> >> The most important questions and answers are at
> >> http://www.dev-archive.com/free/press/2000/vs-qalist.asp
> >
> >Ok lemme explain what I meant. Go to the above page and read
> >these two questions there -
> >
> >Do you know which STL/version is going to be shipped with VC7?
> >Does VC.NET support ANSI/ISO C++? Can VC.NET use Blitz++?
> >
> >Ronald Laeremans of Microsoft doesn't sound convincing at all
> >if you analyze what he says. As answer to the first question
> >he says that we will ship a very conformant version of STL. And
> >in the next one he says that they won't be able to give partial
> >template specialization in VC7. Now probably what he doesn't
> >know is that it is *not* possible to have a conformant STL
> >without that. You see the standard says that you need to give
> >out iterator traits for all iterators for a number of functions
> >to work properly in <algorithm>. Now you *cannot* do that fully
> >without partial template spcialization. That happens coz the
> >only way to give iterator traits for ordinary pointers is by
> >partial template specialization.
> >But most poor clueless VC users don't know anything about that
> >and are being taken for a ride when they say that STL will
> >be conformant. Why promise something you cannot and more impor-
> >tantly *will not* deliver? Most advanced users of C++ have been
> >crying for better template support in VC for a long time now.
> >Doesn't look like coming. Its all so sickening!
> >Thanks,
> >Shiv
>
>

"Antony Andrews" <asa@asa.org> wrote in message
> Have you tried testing if VC7 now supports covariant return types?

Well it doesn't seem to support covariant return types.
Thanks,
Shiv

I just wanted to say that I am working with Shiv to find out why all the feedback
he tried to give never got far enough.

As an indication for other here. I have found out that the ngwssdk@microsoft.com
alias was intended for feedback on SDK documentation errors, not for feedback
on .Net or Visual Studio in general. This fact doesn't seem to have been
communicated very clearly.
Seeing the purpose of this alias, we had not set up a guaranteed reply mechanism
and have this alias staffed by almost exclusively user education people.

I will repeat my offer from the previous message. If readers here have feedback
they want to give directly to the Visual C++ compiler team, feel free to
send it to me directly at mailto:ronlaere@microsoft.com

-Ronald-

"Ronald Laeremans" <ronlaere@microsoft.com> wrote in message
news:39935530$1@news.dev-archive.com...

Ya'know... I really should be better about remembering names.. OK, The
"Ronald" I'm sure about; it's the "Laeremans" that's a bit hazy... Are you
the (former) MVP (of the funny hats)?

--
Truth,
James Curran
http://www.NJTheater.com (Professional)
http://www.NovelTheory.com (Personal)
http://www.BrandsForLess.com (Day Job)

> Have Microsoft opened .NET newsgroups? I don't see them on
> msnews.microsoft.com...

Yes they have. Refresh your news reader. They are definitely there and have
been since Monday.

Robert Scoble
http://www.vslive.com <<-- The Fall's Largest Visual Studio Conference

###

> microsoft.public.net...

Here's the complete list of .NET Frameworks SDK Newsgroups on the Microsoft
NNTP server. If you use Outlook Express you can just click on these.

news://msnews.microsoft.com/microsoft.public.net.framework.aspplus.general
news://msnews.microsoft.com/microsoft.public.net.framework.classes.general
news://msnews.microsoft.com/microsoft.public.net.framework.runtime.general
news://msnews.microsoft.com/microsoft.public.net.framework.sdk.general
news://msnews.microsoft.com/microsoft.public.net.framework.sdk.setup
news://msnews.microsoft.com/microsoft.public.net.general
news://msnews.microsoft.com/microsoft.public.net.vb.general
news://msnews.microsoft.com/microsoft.public.net.framework.vc.general
news://msnews.microsoft.com/microsoft.public.net.framework.net.xml

Or you could just visit http://msdn.microsoft.com/net/#newsgroups which has
links to all the newsgroups too.

Hope that helps.

Robert Scoble
http://www.vslive.com <<-- The Fall's Largest Visual Studio Conferences

###

Rune,

I would visit news://msnews.microsoft.com/microsoft.public.net.vc.general/
and post this there.

Robert Scoble
http://www.vslive.com <<-- The Fall's Largest Visual Studio Conference

###

"Rune Huseby" <rune@huseby.as> wrote in message
news:399AAC25.54F7E7BF@huseby.as...
> Ronald Laeremans wrote:
> >
> > I should really stress that feedback DOES matter. Feedback however is
much
> > more effective when given directly to Microsoft Product Support or one
of
> > our feedback aliases or on one of the Microsoft monitored newsgroups.
>
> Do you have some explicit links/adresses?
>
> You should also provide some kind of feedback when you receive feedback.
> I have reported several bugs/issues through the support pages,
> but I have never ever received any feedback confirming/rejecting
> my reports. Not even in the form of new KB articles adressing the issue.
>
> --
> Rune H. H. Huseby

On 9 Aug 2000 19:30:07 -0700, "Ronald Laeremans" <ronlaere@microsoft.com>
wrote:

} We did not have the resources for VC.Net 7.0 to get to the few major areas
} in template support where Visual C++ has no real support. The 3 you mention
} below are in that category. Since you are posting in a newsgroup about .Net,
} you should probably have a good idea on where we did spend a majority of
} our resources in this release.

I see.
We develop mupltiplatform software and were struggling with these issues
long enough. We will switch compilers as soon as possible.
There will definitely will no upgrade to VC7 for us.

} All the members of the C++ team consider standards conformance a very important
} issue, and we don't like the fact that we couldn't do more in this release.
} Planning is getting underway for the next version of Visual C++ and adding
} the missing pieces is a top issue on the agenda.

Sorry, we've waited long enough.

Best regards,
Alex.