Using "numeric_limits<double>::min()" in MFC application

[Jalal]

I am trying to use numeric_limits<double>::min() in an MFC application
in Microsoft Visual C++.NET 2003. I am having some difficulties here.
The following are the parts of a simple program I wrote and
corresponding error massages. I am confused here. Similar codes work
fine in Win32 Console Project. Can you please help me to resolve
this? Thank you for your helps.

Best Regards,

Jalal



Code
-------------------------------
// Created by Microsoft Visual C++.NET
#include "stdafx.h"
#include "test.h"
#include "testDlg.h"
#include ".\testdlg.h"

// I have added to use numeric_limits
#include <limits>
using namespace std;


void CtestDlg::OnBnClickedSubmit()
{
double y;

// TODO: Add your control notification handler code here
UpdateData(TRUE);

y = numeric_limits<double>::min();

// Writing y in dialog box
x.Format("%.10f", double(y));

UpdateData(FALSE);
}


Error massage:

 

[Pete Becker]

warning C4003: not enough actual parameters for macro 'min'
error C2589: '(' : illegal token on right side of '::'
error C2059: syntax error : '::'

The windows headers do you the favor of providing macros named 'min' and
'max' that can screw up your code. To get rid of 'em, add

#define NOMINMAX

to your stdafx.h file, before any #include directives.

 

[Alf P. Steinbach]

* (e-mail address removed) (Jalal) schriebt:

I am trying to use numeric_limits<double>::min()
Yes?


in an MFC application in Microsoft Visual C++.NET 2003.

That's off-topic here, but you'll get answers to such questions in
e.g. [comp.os.ms-windows.programmer.win32] (I might even answer it
there myself ;-) ), or in a Microsoft newsgroup.

I am having some difficulties here.
The following are the parts of a simple program I wrote and
corresponding error massages. I am confused here. Similar codes work
fine in Win32 Console Project. Can you please help me to resolve
this? Thank you for your helps.

To top it off you have not included the _relevant_ parts of your program.
Please do that when reposting your question in e.g. the group mentioned
above. Might also be a good idea to state that the question was earlier
posted in [comp.lang.c++] but off-topic there.

 

[Rolf Magnus]

* (e-mail address removed) (Jalal) schriebt:

Yes?

What makes you doubt that?

in an MFC application in Microsoft Visual C++.NET 2003.

That's off-topic here, but you'll get answers to such questions in
e.g. [comp.os.ms-windows.programmer.win32] (I might even answer it
there myself ;-) ), or in a Microsoft newsgroup.

The question isn't actually about MFC, though the problem itself seems
to be related to the windows platform. However, the OP probably didn't
know that. He just got an error message about

To top it off you have not included the _relevant_ parts of your
program.

He hasn't? What would be missing?

Please do that when reposting your question in e.g. the group
mentioned above. Might also be a good idea to state that the question
was earlier posted in [comp.lang.c++] but off-topic there.

The question wasn't, even if the answer might be.

 

[Alf P. Steinbach]

What makes you doubt that?

Why do you think I doubt what?

in an MFC application in Microsoft Visual C++.NET 2003.

That's off-topic here, but you'll get answers to such questions in
e.g. [comp.os.ms-windows.programmer.win32] (I might even answer it
there myself ;-) ), or in a Microsoft newsgroup.

The question isn't actually about MFC, though the problem itself seems
to be related to the windows platform.

It is about one particular vendor's header files for the Windows platform.

However, the OP probably didn't
know that. He just got an error message about
numeric_limits<double>::max(), so he wanted to know what he's doing
wrong.

One thing he or she did wrong was to not try it out in standard C++ before
asking in a standard C++ Usenet group.

He hasn't? What would be missing?

What was missing was the include file where the vendor-specific header
files were erronously included for C++, namely [stdafx.h].

Please do that when reposting your question in e.g. the group
mentioned above. Might also be a good idea to state that the question
was earlier posted in [comp.lang.c++] but off-topic there.

The question wasn't, even if the answer might be.

Both the question and the answer were off-topic. The question because it
concerned use of one particular vendor's header files on one particular
platform, as could easily be ascertained by trying standard C++. The answer
for the same reason, but then it's toss-of-the-coin whether to give a very
short "satisfying" answer (short-term) or point to a more appropriate group.

 

[P.J. Plauger]

Both the question and the answer were off-topic. The question because it
concerned use of one particular vendor's header files on one particular
platform, as could easily be ascertained by trying standard C++. The answer
for the same reason, but then it's toss-of-the-coin whether to give a very
short "satisfying" answer (short-term) or point to a more appropriate

group.

Gosh. And I just tried compiling a simple program with gcc.
The headers look reasonably conforming, and yet I get
this diagnostic from the linker that sqrt is undefined.
Had I known that I failed to write the -lm flag on the
compiler command line, I would have known *not* to ask
for help from this newsgroup (where many contributors
happen to know the answer), but to go to a gcc specific
newsgroup instead. Where's prescience when you really
need it?

The C++ Standard is a worthless document without real
live implementations of the language it describes.
Sadly, there are no standards for the kind of gotchas
different implementations inflict on us. If only there
were a newsgroup that understood this, or at least had
less zealous Topic Police.

P.J. Plauger
Dinkumware, Ltd.
http://www.dinkumware.com

 

[Alf P. Steinbach]

group.

Gosh. And I just tried compiling a simple program with gcc.
The headers look reasonably conforming, and yet I get
this diagnostic from the linker that sqrt is undefined.
Had I known that I failed to write the -lm flag on the
compiler command line, I would have known *not* to ask
for help from this newsgroup (where many contributors
happen to know the answer), but to go to a gcc specific
newsgroup instead.

Well you're so experienced by now, having written all these libraries and
whatnot, that you know that a linker issue is pure tool usage and is
off-topic in this group. The same goes for how to include <windows.h>
properly for C++, or how to use MS Visual Studio. This is in the FAQ.

I don't understand why the heck you're jumping in here, because you sure
as h... do know all this, and would never be confused by a linker issue.

Where's prescience when you really need it?

Indeed, one cannot know in advance the details of why an error occurs.

But one can know that it is to do with linking (pure tool usage).

Or that is to do with #include of Microsoft's <windows.h> (pure MS issue).

The C++ Standard is a worthless document without real
live implementations of the language it describes.
Sadly, there are no standards for the kind of gotchas
different implementations inflict on us. If only there
were a newsgroup that understood this, or at least had
less zealous Topic Police.

I agree with that. Such groups exist. E.g., [no.it.programmering.c++],
which is a Norwegian group where English threads are supported -- sometimes
frequented by Bjarne, and perhaps in the future also you?

Also, perhaps what's regarded as on-topic here should be loosened up a bit.

I've argued for that before, and will probably also in the future...

 

[Julie]

:
[snip]

Also, perhaps what's regarded as on-topic here should be loosened up a bit.

Funny thing is that since this newsgroup doesn't have a charter, there really
isn't a notion of on or off topic here...

The FAQ ain't it, either.

First things first, let's get a charter for the newsgroup.

 

[E. Robert Tisdale]

Alf P. Steinbach wrote:
[snip]

Also, perhaps what's regarded as on-topic here should be loosened up a bit.

Funny thing is that since this newsgroup doesn't have a charter,
there really isn't a notion of on or off topic here...

There certainly *is* strong opinion
about what is on and off topic in the comp.lang.c++ newsgroup.
Whether there is consensus or not is independent
of whether there is a charter or not.

The FAQ ain't it, either.

First things first, let's get a charter for the newsgroup.

A charter won't help build consensus.
A charter won't help enforce rules.
A charter probably won't even help end discussion
about what is on or off topic in this newsgroup.

What helps is if you ignore off-topic articles.
If a subject is off-topic in the comp.lang.c++ newsgroup,
it is usually sufficient to simply mention that fact
to alert unwary subscribers who might otherwise respond.

Almost all off-topic articles are posted by regular subscribers
as an off-topic digression from a legitimate thread.
Only a tiny fraction are posted by new subscribers.

The comp.lang.c++ has a higher signal-to-noise ratio
than newsgroups that have a charter.
Let's not screw that up.

 

[Alf P. Steinbach]

:
[snip]

Also, perhaps what's regarded as on-topic here should be loosened up a bit.

Funny thing is that since this newsgroup doesn't have a charter, there really
isn't a notion of on or off topic here...

The FAQ ain't it, either.

First things first, let's get a charter for the newsgroup.

Well, I'm not sure of the technicalities of retrofitting a charter, but in
principle I agree with you -- the infamous "someone" should do this!

The FAQ as well as the monthly posting should of course include the charter,
when and if it gets written and passes voting (caveat: procedure for that?).

A good place to start might be the [comp.lang.c++.moderated] charter, e.g. at
<url:
ftp://ftp.isc.org/usenet/news.announce.newgroups/comp/comp.lang.c++.moderated>:

<CLC++M CHARTER>
comp.lang.c++.moderated is a moderated news group for discussion of common
C++ programming issues. Welcomed topics of discussion will include any topic
directly related to the C++ programming language. The moderators will decide
which posts are of general interest to the worldwide C++ community. The
moderators are free to reject any post which is deemed to be system-specific,
covered by the comp.lang.c++ FAQ(s), and/or more appropriate to another
Usenet discussion group. The following is a short list of posts which
will be considered off-topic by the moderators.

- System-specific problems which are better addressed in existing
system-specific newsgroups may be considered off-topic. (Here we
are using the term "system-specific" in reference to a particular
computer system configuration.)
- Questions answered in the comp.lang.c++ FAQ(s) will be considered
off-topic.
- Obvious homework problems and trivial questions that can be answered by
quick inspection of any C++ text may be considered off-topic.
- Very broad questions which could equally apply to any computer programming
language may be considered off-topic.
- Advocacy and "flame wars" about C++ will be considered off-topic.
- Advertising shall be considered off-topic.
- C++ product announcements shall be at the discretion of the
moderators.

Overall, the moderators shall lean towards accepting posts and the free
exchange of ideas and general discussion of C++ issues.
</CLC++M CHARTER>

What's a bit lacking here, in my opinion, is topicality of library-specific
problems. A library may be system independent (such as Boost), or very widely
used. But there is a difference discussing the use of e.g. boost::shared_ptr
for pimpl idiom, and on the other hand "how do I implement Tiny Basic using
Boost Spirit?" -- which I'd regard as off-topic no matter Boost or not. In
short there must be some general C++ relevance, not just a C++ association.
The charter says any topic directly related to the C++ language is welcomed,
but it doesn't say any topic only vaguely associated to C++ is unwelcome.

Also, there is the question of actually using C++ for those tasks the language
is intended for.

For example, currently the use of C++ to provide bindings for other languages,
such as Java JNI, is off-topic in just about any group except
[comp.programming], but I think such actual practice should be on-topic here.

 

[Julie]

Alf P. Steinbach wrote:
[snip]

Also, perhaps what's regarded as on-topic here should be loosened up a bit.

Funny thing is that since this newsgroup doesn't have a charter,
there really isn't a notion of on or off topic here...

There certainly *is* strong opinion
about what is on and off topic in the comp.lang.c++ newsgroup.
Whether there is consensus or not is independent
of whether there is a charter or not.

The FAQ ain't it, either.

First things first, let's get a charter for the newsgroup.

A charter won't help build consensus.
A charter won't help enforce rules.
A charter probably won't even help end discussion
about what is on or off topic in this newsgroup.

All true, however it doesn't change the fact that a charter should be the first
order of business if better defining newsgroup topicality is needed.

The comp.lang.c++ has a higher signal-to-noise ratio
than newsgroups that have a charter.
Let's not screw that up.

Are you concluding that necessarily having a charter decreases signal to noise?

 

[Ioannis Vranos]

:
[snip]

Also, perhaps what's regarded as on-topic here should be loosened up a

bit.

Funny thing is that since this newsgroup doesn't have a charter, there really
isn't a notion of on or off topic here...

The FAQ ain't it, either.

First things first, let's get a charter for the newsgroup.

But Usenet is still governed by the old days spirit of freedom. That means
that when you think something does not fit to be discussed here, you notify
the user to find a more appropriate newsgroup.






Regards,

Ioannis Vranos

 

[Ioannis Vranos]

All true, however it doesn't change the fact that a charter should be the first
order of business if better defining newsgroup topicality is needed.

The topicality is well known. ISO C++.






Regards,

Ioannis Vranos

 

[Jalal]

Dear Pete:

I have implemented what you have suggested. My program is now
working. Thank you for your helps.

Best Regards,

Jalal