_BLOCK_TYPE_IS_VALID error

Hi,

I get an unexpected error when debugging my program which is:

Debug Assertion failed: dbgdel.cpp line:52 Expression: _BLOCK_TYPE_IS_VALID

I searched the net but i couldn't find the solution which applies to my condition.
Here's the part i which think contains the error:

Code:

char* name;
DATA_STRUCT curData;
int id;

id=SendMessage(hListBox,LB_GETCURSEL,NULL,NULL);

name=new char[SendMessage(hListBox,LB_GETTEXTLEN,(WPARAM)id,NULL)+1];
						
SendMessage(hListBox,LB_GETTEXT,(WPARAM)id,(LPARAM)name);
						
curData=ReadData(name,gSelType);
						
SetWindowText(hEditName,curData.name.c_str());
						
SetWindowText(hEditInfo,curData.info.c_str());
						
delete [] name;

Code:

DATA_STRUCT ReadData(char* name,int type)
{
	FILE *file;
	char dir[200]="Data/";
	DATA_STRUCT rc;

	strcat_s(dir,200,typeItems[type]);
	strcat_s(dir,200,"/");
	strcat_s(dir,200, name);
	strcat_s(dir,200,".dat");

	file=fopen(dir,"r");
	fread(&rc,sizeof(DATA_STRUCT),1,file);
	fclose(file);

	return rc;
}

Code:

typedef struct{
	string name;
	string info;
	}DATA_STRUCT;

i thought of putting this into the windows board, but i'm sure there's sth with the c++ part.

So what line is the same as dbgdel.cpp line:52 ?

Jim

Originally Posted by jimblumberg

So what line is the same as dbgdel.cpp line:52 ?

Jim

None of them...

EDIT: here's dbgdel.cpp, line:52:

Code:

 
 _ASSERTE(_BLOCK_TYPE_IS_VALID(pHead->nBlockUse));

Originally Posted by jimblumberg

So what line is the same as dbgdel.cpp line:52 ?

The one inside the CRT that tries to free the char* that the std::strings contained when they were (presumably) fwritten to the file. OP, look up serialization. fwriting a std::string to the file doesn't save the contents, you need to actually fwrite those using c_str() and length. Alternatively you could drop the C file I/O routines and adopt the shiny C++ fstream ones with their "do the right thing" << and >> operations.

Originally Posted by adeyblue

The one inside the CRT that tries to free the char* that the std::strings contained when they were (presumably) fwritten to the file. OP, look up serialization. fwriting a std::string to the file doesn't save the contents, you need to actually fwrite those using c_str() and length. Alternatively you could drop the C file I/O routines and adopt the shiny C++ fstream ones with their "do the right thing" << and >> operations.

But ReadData works for some of the files, while the others fail... What's the explanation of this?

Originally Posted by new_in_c++

But ReadData works for some of the files, while the others fail... What's the explanation of this?

Chance, e.g., maybe the contents of the strings were sufficiently small that they were stored in a member char array that was copied, or maybe it so happens that the pointers were restored to point to the same memory.

If the data that you write contains pointers, then only the value of the pointer gets printed, not the actual data that is being pointed to. If you then later try to read the file, you will read the pointers, but those pointers are no longer pointing to the data (since the data itself is no longer in memory).

As far as i understand, everything will be solved if i use fstream, right?

What a mish-mash of C and C++. Yuck.

Might I suggest, if you're going to use C I/O, sprintf?

Code:

DATA_STRUCT ReadData(char* name,int type)
{
	FILE *file;
	char dir[200]="Data/";
	DATA_STRUCT rc;

	strcat_s(dir,200,typeItems[type]);
	strcat_s(dir,200,"/");
	strcat_s(dir,200, name);
	strcat_s(dir,200,".dat");

can be replaced with

Code:

DATA_STRUCT ReadData(const char* name,int type)
{
	FILE *file = NULL;
	char dir[PATH_MAX]={ 0 };
	DATA_STRUCT rc = { 0 };

        sprintf(dir, "Data/%s/%s.dat", typeItems[type], name);
        file=fopen(dir,"r");
	if (file) {
            fread(&rc,sizeof(DATA_STRUCT),1,file);
	    fclose(file);
        }
        return rc;

Although I think you should be passing in your DATA_STRUCT and returning a status to your caller, but it seems error checking isn't all that important to you. Not a good trait for a programmer.

You propose to turn a safer solution into an error prone one? If you go entirely c++, you should not have to worry about these problems.

Originally Posted by Elysia

You propose to turn a safer solution into an error prone one? If you go entirely c++, you should not have to worry about these problems.

What do you mean by "entirely c++"?

What's more error-prone about what I provided? Other than I should have used snprintf over sprintf.

Code:

snprintf(dir, sizeof(dir), "Data/%s/%s.dat", typeItems[type], name);

or is it because I didn't use the sainted Windows "safe stdio" calls?

You used sprintf over strcat_s, which would make it less safe.
But why are we bothering with C API in the first place? This is C++. If we want to use sprintf, we can use Boost.Format. Or we can use std::string.

Don't try and do this:

Code:

name=new char[SendMessage(hListBox,LB_GETTEXTLEN,(WPARAM)id,NULL)+1];

Trying to do all that one one line spells trouble. At the very least it means that you're never checking the return value of SendMessage is what you expect.
What if it's returning -1?!
Split it over two lines.