C / C++ / MFC

13-Jul-06 3:27

Hi together,

I tried to compile the fdstream classes from www.josuttis.com with VC6, but didn't succeed. I cut down the code to what doesn't work:

#include <streambuf>
#include <istream>
 
class mybuf : public std::streambuf
{
protected:
	int m_i;
public:
	mybuf(int i) : m_i(i) {}
};
 
class myistream : public std::istream
{
protected:
	mybuf buf;
public:
	myistream(int i) : std::istream(0), buf(i) {}
};
 
int main(int argc, char* argv[])
{
	myistream strm(5);
	return 0;
}

On this code I get the following two errors:

error C2512: 'basic_istream<char,struct std::char_traits<char> >' : no appropriate default constructor available
error C2614: 'myistream' : illegal member initialization: 'istream' is not a base or member

I can't figure out what's the problem. It seems std::istream(0) isn't accepted, but I don't know why. Dead | X|

Can anyone help me?

Thank you very much in advance.

Regards,
Marcus.

Zac Howland13-Jul-06 4:21

Re: Inheritance problem [modified]

13-Jul-06 4:21

khb wrote:
error C2512: 'basic_istream<char,struct std::char_traits<char=""> >' : no appropriate default constructor available

std::istream is a typedef for basic_istream<char, char_traits<char> >. basic_istream has the following constructor:

explicit basic_istream(basic_streambuf<Elem, Tr> *_Strbuf, bool _Isstd = false);

Do you really want your streambuf to be null? If so, try casting 0 as follows:

std::istream((std::streambuf*)0);

If you decide to become a software engineer, you are signing up to have a 1/2" piece of silicon tell you exactly how stupid you really are for 8 hours a day, 5 days a week

Zac

khb13-Jul-06 5:13

13-Jul-06 5:13

Yeah, the streambuf should to be null, as in the original implementation it is replaced by an instance of the inherited streambuffer class using rdbuf().

However, casting 0 to std::streambuf* results in the same errors.

If you have another idea on how I get the istream constructor called correctly, I would be really happy.

Thank you very much for your help,
Marcus.

Zac Howland13-Jul-06 5:18

13-Jul-06 5:18

Strange. Not sure why that would be a problem ... I was able to compile your code using g++ with no problems.

If you decide to become a software engineer, you are signing up to have a 1/2" piece of silicon tell you exactly how stupid you really are for 8 hours a day, 5 days a week

Zac

earl13-Jul-06 5:20

earl

13-Jul-06 5:20

The original author is using VC6.

For the love of god, VS2005 Express is free. Get it.

earl

Zac Howland13-Jul-06 5:25

13-Jul-06 5:25

I realize that. Generally, GNU is more strick about things like this that Microsoft's compiler.

I have VS2005 Pro ... but it doesn't run on Linux, which is what I have to use at work.

If you decide to become a software engineer, you are signing up to have a 1/2" piece of silicon tell you exactly how stupid you really are for 8 hours a day, 5 days a week

Zac

earl13-Jul-06 5:29

earl

13-Jul-06 5:29

Sorry Zac, I meant to reply to the original author.

VS6 is broken in some ways; I just think that trying to make things work on it is kind of silly when you have a far superior alternative available for free. And the gnu compiler chain is quite nice, but the 3.x series is far superior to VS6 in terms of the probabilty of standard-conforming code compiling and working.

earl

Zac Howland13-Jul-06 5:33

13-Jul-06 5:33

earl wrote:
And the gnu compiler chain is quite nice, but the 3.x series is far superior to VS6 in terms of the probabilty of standard-conforming code compiling and working.

Agreed. But sadly, there are many companies that still do all their development in VS6.

If the OP hasn't already done so, he might want to download the fixed STL from Dinkumware's site (they are the ones that wrote the version that ships with VS). There were several bugs that were fixed, not sure if this might be one of them.

If you decide to become a software engineer, you are signing up to have a 1/2" piece of silicon tell you exactly how stupid you really are for 8 hours a day, 5 days a week

Zac

khb13-Jul-06 8:01

13-Jul-06 8:01

Unfortunately, I have to stick to VS6. It's really bad that the behavior might come from bugs in the STL Frown | :(

Can I simply replace the problematic STL classes? If so, can you provide the Link to the Dinkumware's site?

Thank you for you discussion. That brought some light into the dark. I already started thinking that I'm getting to old for understanding C++ Wink | ;)

Therefore I like your slogan, Zac!

Regards,
Marcus.

Zac Howland13-Jul-06 8:03

13-Jul-06 8:03

I think Microsoft has a link in MSDN for a service pack download that does it. But here is Dinkumware's site: http://www.dinkumware.com/.

If you decide to become a software engineer, you are signing up to have a 1/2" piece of silicon tell you exactly how stupid you really are for 8 hours a day, 5 days a week

Zac

khb13-Jul-06 9:00

13-Jul-06 9:00

Thanks a million!

Regards,
Marcus.

How to setrange of spin control

earl13-Jul-06 12:04

earl

13-Jul-06 12:04

Marcus,

There is simply quite a bit of valid code that the VS6 compiler is unable to compile. This in turn limits the quality of the STL that can be provided for it. At the end of the day, it's an 8 year old compiler that was written before the C++ standard was finalized IIRC.

earl

vc++_fragrance13-Jul-06 2:06

vc++_fragrance

13-Jul-06 2:06

Hi,
In my project i have to initially display the data from database and also i have to provide the facility to the user to change the value. I have done it in onitdialog message handler where the recordset operations have already handled.
i compiled it successfully but while running i got the assertion saying that

assertion in .exe file
File:afxcmn.inl
line:375

plz help me.
Thanks in advance.

tejaswini

*Dreamz13-Jul-06 2:25

*Dreamz

13-Jul-06 2:25

Can you please make it a little more clear what exactly you did in OnInitDialog() ?

vc++_fragrance13-Jul-06 2:33

vc++_fragrance

13-Jul-06 2:33

BOOL CCmpDataEntryDlg::OnInitDialog() //CmpDataEntryDlg is my propertysheet
{
m_SpinDropRate.SetRange(5,19); //to set the range for spin control

m_pSet = new CCampaignSet(); //this is my recordset
if(!m_pSet->IsOpen())
{
CWaitCursor wait;
m_pSet->Open();

}

CPropertyPage::OnInitDialog();
UpdateData(FALSE);

return TRUE; // return TRUE unless you set the focus to a control
// EXCEPTION: OCX Property Pages should return FALSE
}

David Crow13-Jul-06 2:48

David Crow

13-Jul-06 2:48

tejaswini.g wrote:
CPropertyPage::OnInitDialog();

This statement must be called first (otherwise the controls do not get created).

"The largest fire starts but with the smallest spark." - David Crow

"Judge not by the eye but by the heart." - Native American Proverb