C / C++ / MFC

Solved - gdi+ bitmap save on existing file - showing error

24-Dec-20 23:33

Your file mImagePath1 is still open to the Bitmap object. You should use a different path to save the new image.

Gopi Nath24-Dec-20 23:53

Gopi Nath

24-Dec-20 23:53

Thank you Richard.

This helps me to solve my issue.

Regards,
Gopi.

std:strstream does not work

Haakon S.22-Dec-20 2:38

22-Dec-20 2:38

In Visual C++ 2012 I have written this code:

strstream& operator>>(strstream& Stream,CComponent* pCmp){
..code
}

std::strstream read_file((char*)filbuf,file_size,ios_base::out);

pComp=new CComponent;
read_file>>pComp;

This works perfectly. But when I run it on my newly installed Visual C++ 2019 it compiles all right, but the output is wrong.

I see that std::strstream is deprecated, but it should still function. If I try to put std::stringstream instead, it will not accept the >> operator.

Richard MacCutchan22-Dec-20 4:14

22-Dec-20 4:14

Haakon S. wrote:
it compiles all right, but the output is wrong.

You need to explain what that means.

Haakon S.22-Dec-20 22:21

22-Dec-20 22:21

Ok, some details. Everything seems to be identical when I check the debug data. The text file is referenced in the same way by filebuf.

The first member of CComponent is specified as WORD m_nNo and the this value in the first component is 56.

I put some test code on top of my function to try to find out what is going on.

std::strstream& operator>>(std::strstream& Stream,CComponent* pCmp){
	char txt[4];
	Stream >> txt;
.
.
.
}

The watch window displays the following output from the new C++ version:

- txt 0x00cfd5fc "" char[4]
[0] 0 '\0' char
[1] -52 'Ì' char
[2] -52 'Ì' char
[3] -52 'Ì' char

From my older version I get the correct output:

- txt 0x0055e25c "56" char[4]
[0] 53'5' char
[1] 54'6' char
[2] 0 '\0' char
[3] -52 'Ì' char

It seems that the extraction operator >> simply does not work anymore. In that case, what is the best alternative for reading data from a text file. If I replace strstream with stringstream I get the message that stringstream does not have a >> operator.

modified 23-Dec-20 4:35am.

Richard MacCutchan22-Dec-20 23:37

22-Dec-20 23:37

If you are reading from a file then you should be using one of the fstream classes.

Stefan_Lang23-Dec-20 23:59

Stefan_Lang

23-Dec-20 23:59

Looks to me like it is the *old* version that wasn't working: your code overwrites your text variable, so why would you expect it to still hold the old value? Or maybe it is just that the VS2012 debugger didn't show these contents properly because it didn't recognize that you were writing to it.

GOTOs are a bit like wire coat hangers: they tend to breed in the darkness, such that where there once were few, eventually there are many, and the program's architecture collapses beneath them. (Fran Poretto)

Stefan_Lang23-Dec-20 23:56

Stefan_Lang

23-Dec-20 23:56

First, stringstream is not strstream. You may need to specifically use ostringstream instead. But anyway, like Richard already stated, you should be using fstream, or ofstream.

Haakon S.24-Dec-20 0:37

24-Dec-20 0:37

Hi.

Thanks for helping.

ifstream accepts the >> operator and get() function.

std::ifstream read_file((char*)filbuf,file_size,ios_base::out);
.
.
 pComp=new CComponent;
 read_file>>pComp;

This function is called:

ifstream& operator>>(ifstream& Stream,CComponent* pCmp){

Stream>>pCmp->m_nNo;
Stream.get(pCmp->m_formula,15);
.
.
}

Everything works fine, except that nothing is actually read.

Regards, Haakon.

Richard MacCutchan24-Dec-20 4:18

24-Dec-20 4:18

Haakon S. wrote:
Everything works fine, except that nothing is actually read

Well both of those statements cannot be true.

Stefan_Lang24-Dec-20 5:10

Stefan_Lang

24-Dec-20 5:10

Ah, yes, you're right, of course I meant ifstream, not ofstream. Wink | ;)

But this statement shouldn't compile:

std::ifstream read_file((char*)filbuf,file_size,ios_base::out);

See ifstream::ifstream - C++ Reference[^] : none of the constructors match the one you're calling. Specifically no constructor does have three arguments, and none expect anything like a file size. (and why should they?)

Furthermore, I wonder what that variable filebuf is, and why it needs to be casted to char* ? A typecast is almost always an indication of an incorrect implementation, and in this case I think you are passing the wrong parameter! The first parameter, if you pass any, should always be the file name. What is the name of the file you want to read? You should pass it here.

Haakon S.24-Dec-20 11:59

Re: std:strstream does not work. Solution (sort of)

24-Dec-20 11:59

Hi.

It does compile. But you are right. When the read file is invoked, it does not take filbuf along with it, I can see that in the debugger. I have to figure that out.

Anyway, thanks alot for your hjelp. Have a Merry Christmas. During the holiday I will try to fix the problem. If I do, I will report back.

Best regards, Haakon.

Haakon S.26-Dec-20 1:11

Re: std:strstream does not work. Solution (sort of)

26-Dec-20 1:11

Hi.

I have managed to make things work, albeit not in the way I want to.

The data to be read is in a file called datf.txt. Having this file in the same directory as the exe file and simply writing

std::ifstream read_file("datf.txt");

is working. However, I want to have the data incorporated as part of the exe file. I have therefore included the CResourceTextFile class by Hans Dietrich. In the progarm resource:

IDU_COMP_FILE           TEXT                    "datf.txt"

and further

CResourceTextFile rf;
rf.Open(NULL, _T("IDU_COMP_FILE"));
TCHAR* filbuf=rf.GetTextBuffer();
std::ifstream read_file(filbuf);

does not work anymore. It did for sstream in the older version of C++ but not for ifstream in the latest.

Randor 26-Dec-20 3:28

Randor

26-Dec-20 3:28

I think your original idea in your first post... std::stringstream was probably the best way.

You just needed to add a istream operator >> to handle the CComponent. Something like this:

C++

class CComponent
{
private:
	std::string something;
public:
	CComponent() noexcept = default;
	friend std::istream& operator >> (std::istream& in, CComponent& component);
};

istream& operator >> (istream& in, CComponent& c)
{
	in >> component.something;
	return in;
}

Which could then be used like this:

C++

const std::string test_haakon = "Lorem ipsum dolor sit amet";
CComponent test_component;
std::stringstream test_string_stream(test_haakon.c_str());
test_string_stream >> test_component;

Merry Christmas to you and your family.

Best Wishes,
-David Delaune

UPDATE:
I really hate answering these modern C++ questions because it takes a lot more effort. I shouldn't have said that it was 'probably the best way' because if you use the code above it will make a copy of the data three times.

If you are able to use C++17 in your project then you will be able to avoid one of those copies by deriving a class from std::basic_streambuf as such:

C++

class CComponent
{
private:
	std::string something;
public:
	CComponent() noexcept = default;
	friend std::istream& operator >> (std::istream& in, CComponent& component);
};

template<typename T = CHAR, typename CT = std::char_traits<T> >
class make_stream_buffer_no_copy : public std::basic_streambuf<T, CT>
{
public:
	make_stream_buffer_no_copy(CHAR* p, size_t len)
	{
		basic_streambuf<T>::setg(p, p, p + len);
		basic_streambuf<T>::setp(p, p + len);
	}

	std::basic_stringstream<T> get_stream()
	{
		return std::basic_stringstream<T>(basic_streambuf<T>::pbase(), ios_base::in | ios_base::out);
	}
};

istream& operator >> (istream& in, CComponent& c)
{
	in >> c.something;
	return in;
}

Which you would use like this:

C++

CHAR p2[] = "Lorem ipsum dolor sit amet";
CComponent test_component2;
make_stream_buffer_no_copy no_copy(p2,strlen(p2));
no_copy.get_stream() >> test_component2;

You can avoid ALL copying if you are willing to extend your CComponent class.

modified 26-Dec-20 11:03am.

Configure Script Issue

jblixt21-Dec-20 12:37

jblixt

21-Dec-20 12:37

This was the most appropriate place that I saw to post a C build system question that I saw so, sorry if there is a better place that I missed. Anyhow, I have run into an issue running configure where the linker is not finding the library that I'm specifying. I'm down to guesses so, any direction would be helpful. The configure command that I've constructed is below but, basically, I'm using LDFLAGS to point to the .so file that resides in a non-standard location. I'm using CFLAGS to point to the includes directory where the header files reside and LIBS to indicate the library that I'm using. There is another directory /usr/local/bin that contains three binaries that I don't know how to specify with the configure command. I'm trying to build GNU TLS from source after installing Nettle from the source files as well. Thank you in advance!

./configure CFLAGS="-I/usr/local/include/nettle/" LDFLAGS="-L/usr/local/lib64" LIBS="-lnettle"

Re: Configure Script Issue

Randor 21-Dec-20 13:01

Randor

21-Dec-20 13:01

Hi,

I think on Linux the LD_LIBRARY_PATH environment variable is used to find shared libs. You might be looking for ldconfig and /etc/ld.so.conf if you are on Linux.

Re: Configure Script Issue

k505421-Dec-20 17:42

k5054

21-Dec-20 17:42

You don't say whether make completes successfully or not. The compiler flags given tell the compiler where to find the pieces to buld the program. If the compile has completed successfully, but when you try to run the program you get a message

"error while loading shared libraries: <libname.so>: cannot open shared object file: No such file or directory

then the problem is at runtime, not compile time. As Randor notes, you can tell the link-loader where to find the library with LD_LIBRARY_PATH

k5054@localhost$ LD_LIBRARY_PATH=/usr/local/lib64 myprog arg1 arg2

sets LD_LIBRARY_PATH for the given command. To set for a single session do

k5054@localhost$ export LD_LIBRARY_PATH=/usr/local/lib64
k5054@localhost$ myprog arg1 arg2

You can add that to your shell's .profile file. and it will be set every time you log in. If you want to set this up permanently for all users ont the system, then as root create a file /etc/ld.so.conf.d/local.conf containing the single line

/usr/local/lib64

Now, as root, run ldconfig, and you should be able to run your program from any login, without needing to set LD_LIBRARY_PATH in your .profile or per session/per command.

you can check what libraries are being loaded using the ldd command

<
k5054@localhost$ ldd /bin/bash
        linux-vdso.so.1 (0x00007ffebf5f6000)
        libtinfo.so.6 => /lib/x86_64-linux-gnu/libtinfo.so.6 (0x00007f7ec6539000)
        libdl.so.2 => /lib/x86_64-linux-gnu/libdl.so.2 (0x00007f7ec6533000)
        libc.so.6 => /lib/x86_64-linux-gnu/libc.so.6 (0x00007f7ec6341000)
        /lib64/ld-linux-x86-64.so.2 (0x00007f7ec66b5000)

If the link-loader can't find a library, it will show

libsomelib.so => not found

Keep Calm and Carry On

Writing a Factory Design Pattern Program with separate Header Files and Class Implementations

Litu Sahoo17-Dec-20 12:26

Litu Sahoo

17-Dec-20 12:26

Hi,

I am trying to learn factory design pattern in c++, writing a program with separate header files and their class implementations in .cpp files. But i am facing problem. I was trying the ShapeFactory from where we implement Circle and Rectangle.

invalid new-expression of abstract class type 'Circle'
invalid new-expression of abstract class type 'Rectangle'

But i am getting these errors and also few other errors. Can someone please help on this.

Regards
Litu

Re: Writing a Factory Design Pattern Program with separate Header Files and Class Implementations

Gerry Schmitz17-Dec-20 19:27

Gerry Schmitz

17-Dec-20 19:27

"new" ... you can't instantiate an "abstract" class; you need to subclass it first or use an existing derived class.

It was only in wine that he laid down no limit for himself, but he did not allow himself to be confused by it.
― Confucian Analects: Rules of Confucius about his food

Re: Writing a Factory Design Pattern Program with separate Header Files and Class Implementations

Richard MacCutchan17-Dec-20 23:36