www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.learn - Access violation using append/concatenate operator "~"

reply V <v pathlink.com> writes:
I have a function that passes a string to be displayed on a screen and 
when using the "~" operator, causes an access violation:

<code>

void printFunc(char[] msg)
{
	msg ~= "\n";	// Error
	SendMessageA(wndConsole, EM_REPLACESEL, 0, cast(LPARAM) &msg);
}

</code>

Am I doing something wrong here?  Thanks :)
Mar 25 2005
parent reply Derek Parnell <derek psych.ward> writes:
On Fri, 25 Mar 2005 19:51:13 -0800, V wrote:


 I have a function that passes a string to be displayed on a screen and 
 when using the "~" operator, causes an access violation:
 
 <code>
 
 void printFunc(char[] msg)
 {
 	msg ~= "\n";	// Error
 	SendMessageA(wndConsole, EM_REPLACESEL, 0, cast(LPARAM) &msg);
 }
 
 </code>
 
 Am I doing something wrong here?  Thanks :)


By any chance is the data being passed a string literal *and* you are
running on a linux system? That would cause one. If so, try recoding it as
...

 void printFunc(char[] msg)
 {
 	SendMessageA(wndConsole, EM_REPLACESEL, 0, cast(LPARAM) (msg~"\n").ptr );
 }

-- 
Derek Parnell
Melbourne, Australia
26/03/2005 4:34:31 PM
Mar 25 2005
next sibling parent reply V <v pathlink.com> writes:
Derek Parnell wrote:

 On Fri, 25 Mar 2005 19:51:13 -0800, V wrote:
 
 

I have a function that passes a string to be displayed on a screen and 
when using the "~" operator, causes an access violation:

<code>

void printFunc(char[] msg)
{
	msg ~= "\n";	// Error
	SendMessageA(wndConsole, EM_REPLACESEL, 0, cast(LPARAM) &msg);
}

</code>

Am I doing something wrong here?  Thanks :)


 
 
 By any chance is the data being passed a string literal *and* you are
 running on a linux system? That would cause one. If so, try recoding it as
 ...
 
  void printFunc(char[] msg)
  {
  	SendMessageA(wndConsole, EM_REPLACESEL, 0, cast(LPARAM) (msg~"\n").ptr );
  }
 


I'm using windows xp and here is the code that precedes the function call:

<code>

char[] text;
GETTEXTEX gtx;
gtx.cb = 151;
gtx.flags = GT_DEFAULT;
gtx.lpDefaultChar = null;
gtx.codepage = CP_ACP;
gtx.lpUsedDefChar = null;
SendMessageA(wndInput, EM_GETTEXTEX, cast(WPARAM)&gtx, cast(LPARAM)&text );
printFunc(text);

</code>

The output is fine if I don't use the append operator but will not start 
on a new line.
Mar 26 2005
next sibling parent reply "John C" <johnch_atms hotmail.com> writes:
"V" <v pathlink.com> wrote in message 
news:d23c7p$26o9$1 digitaldaemon.com...

 Derek Parnell wrote:

 On Fri, 25 Mar 2005 19:51:13 -0800, V wrote:



I have a function that passes a string to be displayed on a screen and 
when using the "~" operator, causes an access violation:

<code>

void printFunc(char[] msg)
{
 msg ~= "\n"; // Error
 SendMessageA(wndConsole, EM_REPLACESEL, 0, cast(LPARAM) &msg);
}

</code>

Am I doing something wrong here?  Thanks :)



 By any chance is the data being passed a string literal *and* you are
 running on a linux system? That would cause one. If so, try recoding it 
 as
 ...

  void printFunc(char[] msg)
  {
  SendMessageA(wndConsole, EM_REPLACESEL, 0, cast(LPARAM) 
 (msg~"\n").ptr );
  }


 I'm using windows xp and here is the code that precedes the function call:

 <code>

 char[] text;
 GETTEXTEX gtx;
 gtx.cb = 151;
 gtx.flags = GT_DEFAULT;
 gtx.lpDefaultChar = null;
 gtx.codepage = CP_ACP;
 gtx.lpUsedDefChar = null;
 SendMessageA(wndInput, EM_GETTEXTEX, cast(WPARAM)&gtx, 
 cast(LPARAM)&text );


You're passing the address of a char[], not a pointer to a null-terminated 
string. Try this:

    char[] text = new char[151];
    ...
    gtx.cb = text.length;
    ...
    SendMessageA(wndInput, EM_GETTEXTEX, cast(WPARAM)&gtx, 
cast(LPARAM)std.string.toStringz(text));


 printFunc(text);

 </code>

 The output is fine if I don't use the append operator but will not start 
 on a new line.
Mar 26 2005
parent V <v pathlink.com> writes:
John C wrote:

 "V" <v pathlink.com> wrote in message 
 news:d23c7p$26o9$1 digitaldaemon.com...
 

Derek Parnell wrote:


On Fri, 25 Mar 2005 19:51:13 -0800, V wrote:




I have a function that passes a string to be displayed on a screen and 
when using the "~" operator, causes an access violation:

<code>

void printFunc(char[] msg)
{
msg ~= "\n"; // Error
SendMessageA(wndConsole, EM_REPLACESEL, 0, cast(LPARAM) &msg);
}

</code>

Am I doing something wrong here?  Thanks :)



By any chance is the data being passed a string literal *and* you are
running on a linux system? That would cause one. If so, try recoding it 
as
...

 void printFunc(char[] msg)
 {
 SendMessageA(wndConsole, EM_REPLACESEL, 0, cast(LPARAM) 
(msg~"\n").ptr );
 }


I'm using windows xp and here is the code that precedes the function call:

<code>

char[] text;
GETTEXTEX gtx;
gtx.cb = 151;
gtx.flags = GT_DEFAULT;
gtx.lpDefaultChar = null;
gtx.codepage = CP_ACP;
gtx.lpUsedDefChar = null;
SendMessageA(wndInput, EM_GETTEXTEX, cast(WPARAM)&gtx, 
cast(LPARAM)&text );


 
 
 You're passing the address of a char[], not a pointer to a null-terminated 
 string. Try this:
 
     char[] text = new char[151];
     ...
     gtx.cb = text.length;
     ...
     SendMessageA(wndInput, EM_GETTEXTEX, cast(WPARAM)&gtx, 
 cast(LPARAM)std.string.toStringz(text));
 
 

printFunc(text);

</code>

The output is fine if I don't use the append operator but will not start 
on a new line. 


 
 
 



Using std.string.toStringsz(text) produces garbage when I display it in 
the console window.
Mar 27 2005
prev sibling parent reply Derek Parnell <derek psych.ward> writes:
On Sat, 26 Mar 2005 01:55:44 -0800, V wrote:


 Derek Parnell wrote:

 On Fri, 25 Mar 2005 19:51:13 -0800, V wrote:
 
 

I have a function that passes a string to be displayed on a screen and 
when using the "~" operator, causes an access violation:

<code>

void printFunc(char[] msg)
{
	msg ~= "\n";	// Error
	SendMessageA(wndConsole, EM_REPLACESEL, 0, cast(LPARAM) &msg);
}

</code>

Am I doing something wrong here?  Thanks :)


 
 
 By any chance is the data being passed a string literal *and* you are
 running on a linux system? That would cause one. If so, try recoding it as
 ...
 
  void printFunc(char[] msg)
  {
  	SendMessageA(wndConsole, EM_REPLACESEL, 0, cast(LPARAM) (msg~"\n").ptr );
  }
 


 
 I'm using windows xp and here is the code that precedes the function call:
 
 <code>
 
 char[] text;
 GETTEXTEX gtx;
 gtx.cb = 151;
 gtx.flags = GT_DEFAULT;
 gtx.lpDefaultChar = null;
 gtx.codepage = CP_ACP;
 gtx.lpUsedDefChar = null;
 SendMessageA(wndInput, EM_GETTEXTEX, cast(WPARAM)&gtx, cast(LPARAM)&text );
 printFunc(text);
 
 </code>
 
 The output is fine if I don't use the append operator but will not start 
 on a new line.


The first thing I notice is that you set the cb field to 151, meaning that
the buffer size is 150 bytes + 1 for the null char. However, you haven't
allocated anything in the buffer.

I think you need "text.length = gtx.cb;" just before sending the
EM_GETTEXTEX message.

Then I think the send message call should more like ...

SendMessageA(wndInput, EM_GETTEXTEX, cast(WPARAM)&gtx, cast(LPARAM)&text[0]
);

In other words, pass the address of the first byte in the buffer. Unlike,
C/C++, when if you declare an array such as char[] text, then use '&text',
you are passing the address of the 8-byte dynamic array descriptor rather
than the address of the first byte in the array. This means that the
SendMessage routine will write over those 8 bytes, and possibly more, thus
messing up your RAM. Access violations are almost guaranteed at some stage.


-- 
Derek Parnell
Melbourne, Australia
http://www.dsource.org/projects/build
26/03/2005 10:27:00 PM
Mar 26 2005
parent V <v pathlink.com> writes:
Derek Parnell wrote:

 On Sat, 26 Mar 2005 01:55:44 -0800, V wrote:
 
 

Derek Parnell wrote:


On Fri, 25 Mar 2005 19:51:13 -0800, V wrote:




I have a function that passes a string to be displayed on a screen and 
when using the "~" operator, causes an access violation:

<code>

void printFunc(char[] msg)
{
	msg ~= "\n";	// Error
	SendMessageA(wndConsole, EM_REPLACESEL, 0, cast(LPARAM) &msg);
}

</code>

Am I doing something wrong here?  Thanks :)



By any chance is the data being passed a string literal *and* you are
running on a linux system? That would cause one. If so, try recoding it as
...

 void printFunc(char[] msg)
 {
 	SendMessageA(wndConsole, EM_REPLACESEL, 0, cast(LPARAM) (msg~"\n").ptr );
 }


I'm using windows xp and here is the code that precedes the function call:

<code>

char[] text;
GETTEXTEX gtx;
gtx.cb = 151;
gtx.flags = GT_DEFAULT;
gtx.lpDefaultChar = null;
gtx.codepage = CP_ACP;
gtx.lpUsedDefChar = null;
SendMessageA(wndInput, EM_GETTEXTEX, cast(WPARAM)&gtx, cast(LPARAM)&text );
printFunc(text);

</code>

The output is fine if I don't use the append operator but will not start 
on a new line.


 
 
 The first thing I notice is that you set the cb field to 151, meaning that
 the buffer size is 150 bytes + 1 for the null char. However, you haven't
 allocated anything in the buffer.
 
 I think you need "text.length = gtx.cb;" just before sending the
 EM_GETTEXTEX message.
 
 Then I think the send message call should more like ...
 
 SendMessageA(wndInput, EM_GETTEXTEX, cast(WPARAM)&gtx, cast(LPARAM)&text[0]
 );
 
 In other words, pass the address of the first byte in the buffer. Unlike,
 C/C++, when if you declare an array such as char[] text, then use '&text',
 you are passing the address of the 8-byte dynamic array descriptor rather
 than the address of the first byte in the array. This means that the
 SendMessage routine will write over those 8 bytes, and possibly more, thus
 messing up your RAM. Access violations are almost guaranteed at some stage.
 
 



Using &text[0] in SendMessageA(wndInput...  works but now the append 
operation is not displayed, meaning the newline is not added, but it 
doesn't throw an exception.  Maybe there is a null terminator at the end 
which is why the newline is not displayed?

<code>

SendMessageA(wndInput, EM_GETTEXTEX, cast(WPARAM)&gtx, 
cast(LPARAM)&text[0]);
text ~= "\n";
printFunc(text);


void printFunc(char[] msg)
{
	SendMessageA(wndConsole, EM_REPLACESEL, 0, cast(LPARAM) 			&msg[0]);
}

</code>
Mar 27 2005
prev sibling parent =?ISO-8859-1?Q?Anders_F_Bj=F6rklund?= <afb algonet.se> writes:
Derek Parnell wrote:


 By any chance is the data being passed a string literal *and* you are
 running on a linux system? That would cause one. 


That is not true. Concatenation to an array creates a new one...

Setting the current array contents by indexing would, however ?

--anders
Mar 27 2005