VC++ Net 2002 - 2003 upgrade

[Jonathan Turkanis]

7.1 is a new broken compiler.

After all, this is supposedly the 8th iteration (or so) of the product.

Of course it's not perfect; no compiler is -- so by your standards all
are broken. By more reasonable standards VC7.1 is first rate.

Jonathan

 

[Bret Pehrson]

I expect a subsequent version to be better and less buggier than the previous.
Microsoft doesn't agree. Look at the history of MSVC. Without fail, the .0
version is extremely buggy and insufficient, the follow on revisions slowly get
better until the next .0 release.

I don't expect perfection in a product, but I do expect consistent progression
(w/o excessive regression).

 

[Hendrik Schober]

[...]

7.1 is a new broken compiler.

[...]
Of course it's not perfect; no compiler is -- so by your standards all
are broken. By more reasonable standards VC7.1 is first rate.

I wouldn't consider a C++ compiler "first
rate" if it fails to do two phase-parsing
on templates. Also, for my taste it ICEs
far to often for beeing first rate.
That isn't to say that VC didn't come a
loooong way since VC6.

Jonathan

Schobi

--
(e-mail address removed) is never read
I'm Schobi at suespammers dot org

"Sometimes compilers are so much more reasonable than people."
Scott Meyers

 

[Guest]

I'd settle for just a COMPLETE product. It seems to me that there were a bunch of stuff that didn't happen to be ready for 7.0 and so they left them out and now we have to buy the whole lot again to get the bits that should have been in there in the first place

I'm not really a sophisticated user. If I get executable C++ code out the end I'm happy. The time it takes to write the code is more important to me than whether it is compiled in one or two phases. That's why I want the forms designer (and the other tools in 7.1).

The other thing that annoys me is the year versioning. 2003 and 2002 sound like 7.0 and 8.0. What's wrong with the old fashioned way.

 

[Bo Persson]

I'd settle for just a COMPLETE product. It seems to me that there were a

bunch of stuff that didn't happen to be ready for 7.0 and so they left them
out and now we have to buy the whole lot again to get the bits that should
have been in there in the first place.

I'm not really a sophisticated user. If I get executable C++ code out the

end I'm happy. The time it takes to write the code is more important to me
than whether it is compiled in one or two phases. That's why I want the
forms designer (and the other tools in 7.1).

The other thing that annoys me is the year versioning. 2003 and 2002 sound

like 7.0 and 8.0. What's wrong with the old fashioned way.

The names were decided by the marketing department, they have no real value.

The C++ compiler calls itself version 13.1 for the 2003 edition. Not that
there has ever been 13 versions...


Bo Persson

 

[Hendrik Schober]

[...]
The names were decided by the marketing department, they have no real value.

They are indeed bad and stupid. How often
is it called "Visual C++ .NET 2003" here
and how often VC7.1? And how often were
people confused between 2002/2003? It's
easy to be confused, compared to 7/8.

The C++ compiler calls itself version 13.1 for the 2003 edition. Not that
there has ever been 13 versions...

I think the compiler started counting with
MSC for DOS.

Bo Persson

Schobi

--
(e-mail address removed) is never read
I'm Schobi at suespammers dot org

"Sometimes compilers are so much more reasonable than people."
Scott Meyers

 

[Jonathan Turkanis]

[...]

7.1 is a new broken compiler.

[...]
Of course it's not perfect; no compiler is -- so by your standards all
are broken. By more reasonable standards VC7.1 is first rate.

I wouldn't consider a C++ compiler "first
rate" if it fails to do two phase-parsing
on templates. Also, for my taste it ICEs
far to often for beeing first rate.
That isn't to say that VC didn't come a
loooong way since VC6.

Okay, I'm not going to argue about the definition of 'first-rate'. ;-)

Let me put it this way. I haven't yet written any tricky
template-based code which I haven't been able to make work correctly
on VC7.1. (I'm not talking about sample programs to test
conformance -- I mean real-world code which I need for some project.)
I can say this only for three other compilers: gcc3.2+, Intel 7.1+ and
como 4.3.3. That puts VC7.1 in the top four in my book.

I'm aware of problems with VC7.1's conformance, but so far they
haven't tripped me up.

Jonathan

 

[Bret Pehrson]

Remember, long ago MSVC was really the C compiler from Lattice...

[...]
The names were decided by the marketing department, they have no real value.

They are indeed bad and stupid. How often
is it called "Visual C++ .NET 2003" here
and how often VC7.1? And how often were
people confused between 2002/2003? It's
easy to be confused, compared to 7/8.

The C++ compiler calls itself version 13.1 for the 2003 edition. Not that
there has ever been 13 versions...

I think the compiler started counting with
MSC for DOS.

Bo Persson

Schobi

--
(e-mail address removed) is never read
I'm Schobi at suespammers dot org

"Sometimes compilers are so much more reasonable than people."
Scott Meyers

 

[Hendrik Schober]

Remember, long ago MSVC was really the C compiler from Lattice...

Was it? I wasn't using MSC back then.

[...]

Schobi

--
(e-mail address removed) is never read
I'm Schobi at suespammers dot org

"Sometimes compilers are so much more reasonable than people."
Scott Meyers

 

[Hendrik Schober]

[...]
Okay, I'm not going to argue about the definition of 'first-rate'. ;-)

Let me put it this way. I haven't yet written any tricky
template-based code which I haven't been able to make work correctly
on VC7.1.

Great. But what does that by itself mean?
I can say, I have written no template code
that I was not able to get work correctly
on VC6. The questions remain: _Why_ didn't
I write it? -- because it wouldn't work --
and How long did it take? -- too long for
too much of the code. [...]

If I'm cornered with different errors spit
out by VC, CW and GCC I usually cook up a
sample and play with Comeau until I got it
done after the book. _Then_ I try to make
the other compilers accept the code. Now,
the interesting thing about this is that,
most of the time, once I got the code the
way it used to be, it is a lot easier to
make the rest of the compilers accept it.
The biggest problem is to get there when
all you have is broken compilers that ICE
on simple syntax errors.
This is where Comeau shines. It is (almost)
100% conforming (so much that I have yet to
find a compiler vendor that wouldn't accept
a bug report if you argue with "but Comeau
does/doesn't accept the code") and it does
produce great error messages.
While you can get your work done with just
about any other compiler, it does take
longer. And it hurts more.

I'm aware of problems with VC7.1's conformance, but so far they
haven't tripped me up.

Missing two-pahes lookup is a PITA if you
write library code. It is stupid to have
your users find silly syntax errors in your
code. I always check my code with CW before
I release it, even if it is meant to be used
with VC only, since CW does check non-
dependend names.

Jonathan

Schobi

--
(e-mail address removed) is never read
I'm Schobi at suespammers dot org

"Sometimes compilers are so much more reasonable than people."
Scott Meyers

 

[Jonathan Turkanis]

[...]
Okay, I'm not going to argue about the definition of 'first-rate'.

;-)

Let me put it this way. I haven't yet written any tricky
template-based code which I haven't been able to make work correctly
on VC7.1.

Great. But what does that by itself mean?
I can say, I have written no template code
that I was not able to get work correctly
on VC6. The questions remain: _Why_ didn't
I write it? -- because it wouldn't work --
and How long did it take? -- too long for
too much of the code. [...]

Okay, that way didn't work. One more try: I've never been forced
to abandon any technique requiring a high degree of
standard-conformance because I couldn't get it to work on VC7.1. This
is in contrast with all previous versions of VC, Borland 5.x, CW8.0
and some previous versions of the other compilers we discussed.

This is where Comeau shines. It is (almost)
100% conforming (so much that I have yet to
find a compiler vendor that wouldn't accept
a bug report if you argue with "but Comeau
does/doesn't accept the code") and it does
produce great error messages.
While you can get your work done with just
about any other compiler, it does take
longer. And it hurts more.

I share your affection for Comeau. I'm not sure I always like its
error messages, though.

Missing two-pahes lookup is a PITA if you
write library code. It is stupid to have
your users find silly syntax errors in your
code. I always check my code with CW before
I release it, even if it is meant to be used
with VC only, since CW does check non-
dependend names.

Fair enough. But I never make silly syntax errors. ;-)

Jonathan

 

[Hendrik Schober]

[...]

Okay, that way didn't work. One more try: I've never been forced
to abandon any technique requiring a high degree of
standard-conformance because I couldn't get it to work on VC7.1. This
is in contrast with all previous versions of VC, Borland 5.x, CW8.0
and some previous versions of the other compilers we discussed.

Yes. I agree with this.

[...]

code. I always check my code with CW before
I release it, even if it is meant to be used
with VC only, since CW does check non-
dependend names.

Fair enough. But I never make silly syntax errors. ;-)

Wow. Need a job?

Jonathan

Schobi

--
(e-mail address removed) is never read
I'm Schobi at suespammers dot org

"Sometimes compilers are so much more reasonable than people."
Scott Meyers

 

[Bo Persson]

Was it? I wasn't using MSC back then.

That's where the confusion started. The Lattice C was version 2.x, so when
MS introduced their first own compiler, it was already version 3.0!

That series ran for a lot of versions, up to 8.0 I guess.

Then they turned Visual, using numbers 1, 1.5, 2, 4, 5, and 6. No version 3
this time!

And it continues.


Bo Persson

 

[Hendrik Schober]

[...]

Was it? I wasn't using MSC back then.

That's where the confusion started. The Lattice C was version 2.x, so when
MS introduced their first own compiler, it was already version 3.0!

That series ran for a lot of versions, up to 8.0 I guess.

Then they turned Visual, using numbers 1, 1.5, 2, 4, 5, and 6. No version 3
this time!

And it continues.

Yeah. Confusion, too.

Bo Persson

Schobi

--
(e-mail address removed) is never read
I'm Schobi at suespammers dot org

"Sometimes compilers are so much more reasonable than people."
Scott Meyers

 

[Ronald Laeremans [MSFT]]

Most likely there will not be a VC Pro edition for Whidbey. Most likely
however the optimizing compiler will be part of VC Standard. None of the 2
is 100% certain yet.

Ronald

 

[Ronald Laeremans [MSFT]]

Hi Nick,

If you look at many products that have low end and full editions in these
price ranges (e.g. Adobe PhotoShop and PhotoShop Elements to give an example
from a very different field), the low end edition does generally not have
upgrade pricing whereas the full edition does.

Ronald

Ron,

Why would I want to pay $US549 for an upgrade for the pro edition when I

only paid about $AUS250 for the std edition? If I'd wanted the pro edition
in the first place I'd have bought it. At that price the MSDN membership
would have also made sense and I would have gotten the upgrade for free,
apparently.

I only want C++. I don't want all the rest. I'm not interested in an

upgrade regardless of price.

 

[Guest]

Ron

You must think that I'm a moron! You are trying to pass off VC++.net as a "low end", cut down or lite version of the whole .net development environment. Where in the name "VC++ .net standard" is the word "lite". How could anyone imagine that C++ is "low end"

There's a difference between a product suite (eg. Office) a bundle (eg. Norton Security products in one box) and a product (Photoshop). The pro version could be called a bundle or suite but you (or your head-up-their-ass marketers) cannot convince me that STANDARD edition is actually a lite edition

I didn't buy a lite version, I don't want a lite version I want C++ with visual studio with all the tools

Microsoft, more than any other company has molded the PC software market. Now it seems that you are changing the rules to suit yourselves. This is the sort of behaviour that ends up in court. (And declining market share

I'm still furious.

 

[Ronald Laeremans [MSFT]]

I am sorry that you are furious, but "Visual C++ Standard Edition"is
definitely what is definitely a "low end" SKU. Just like the same named SKU
was in the 6.0 version. It is in the same price range. If you want to get a
full featured version that one is "Visual Studio Professional Edition". From
VS 7.0 on there are no longer separate Pro versions for each of the
languages.

Ronald

Ron,

You must think that I'm a moron! You are trying to pass off VC++.net as a

"low end", cut down or lite version of the whole .net development
environment. Where in the name "VC++ .net standard" is the word "lite". How
could anyone imagine that C++ is "low end".

There's a difference between a product suite (eg. Office) a bundle (eg.

Norton Security products in one box) and a product (Photoshop). The pro
version could be called a bundle or suite but you (or your head-up-their-ass
marketers) cannot convince me that STANDARD edition is actually a lite
edition.

I didn't buy a lite version, I don't want a lite version I want C++ with

visual studio with all the tools.

Microsoft, more than any other company has molded the PC software market.

Now it seems that you are changing the rules to suit yourselves. This is the
sort of behaviour that ends up in court. (And declining market share)