|
I have a cDialog class that I spawn with new MyClass in another class (MainClass), and I push_back this m_pMyClass pointer on a vector. Now from another location I wanted to delete and zero out the pointer and I tried:
delete myGlobalVector[i];
myGlobalVector[i]= 0;
This didnt seem to do it.
So then I did,
delete MainClass.m_pMyClass;
MainClass.m_pMyClass = 0;
and this worked.
I am puzzled. I thought the first approach would work too..............??
Thanks,
ns
|
|
|
|
|
How does the first solution not work?
Kuphryn
|
|
|
|
|
After I do the delete and NULLing I have a test:
if (!m_pMyCLass)m_button.EnableWindow(TRUE);
but my button doesnt get enabled if I delete it the first (vector) way. The debuuger tells me that indeed the pointer is not zero at this point!
Thanks for helping.
|
|
|
|
|
ns wrote:
if (!m_pMyCLass)m_button.EnableWindow(TRUE);
Please show your actual code.
What you did write here does not use your vector at all!
Who is 'General Failure'? And why is he reading my harddisk?!?
|
|
|
|
|
I push_back m_pDLGSettings onto the vector. Then I get rid of it as below . Then I test it:
void CImageDisplay::OnBsettings()
{
m_pDlgSettings = new CDlgSettings ;
if (m_pDlgSettings == NULL) return;
BOOL ret = m_pDlgSettings->Create(IDD_DLGSETTINGS, this);
m_pDlgSettings->ShowWindow(SW_SHOW);
pDlgSettingsVector.push_back(m_pDlgSettings);
m_bSettings.EnableWindow(FALSE);
}
In another place where I want all windows to destroy and zero out:
if( pDlgSettingsVector.size())
{
for (int i =0; i < pDlgSettingsVector.size(); i++)
{
pDlgSettingsVector[i]->DestroyWindow();
delete pDlgSettingsVector[i];
pDlgSettingsVector[i]=0;
}
pDlgSettingsVector.clear();
}
and in a function of the class that the index [i] refers to:
if (!m_pDlgSettings)
{
m_bSettings.EnableWindow(TRUE);
}
I didnt use the vector here because I dont have the location of the pointer in that vector at this point in code. Instead I access it directly, feeling assured that Ithough deleted it using an index, I can always use it by its name instead of its position in the vector. (m_pDLgSettings is a member). In the place I want to check it (last snippet) I am not using its vector position but its actual variable that I pushed onto the vector..........
Hope I am not hopelessly unclear....
Thanks,
ns
|
|
|
|
|
The item in the vector is actually a copy of the pointer. In other words, you have two pointers to the CDlgSettings object, m_pDlgSettings and pDlgSettingsVector[0].
While push_back() does pass the object by reference, it makes a copy of the object (in this case a CDlgSettings*) to place in the vector. Thus you are setting one pointer to the CDlgSettigns object to NULL, but the other one still points to the now deleted object.
Possible work around - make the vector contain pointers to pointers, thus the vector will contain a pointer to m_pDlgSettings.
Hope this helps,
Tim
|
|
|
|
|
Thanks!!!
Thats a subtle point I was unaware of. Appreciate the workaround suggestion. I am trying it out. Not quite sure what I'll be deleting if I delete the vector element. Its a pointer to a pointer so do I delete *(pDlgSettingsVector[i]) ? i.e. the contents of the pointer contained in the vector.
Thanks,
ns
|
|
|
|
|
No, No, No..
Don't do this. Do not store the pointer returned by new anywhere else except the vector.
All access to that object should be through the pointer stored in the vector. Your problem
here is that you have many copies of the pointers floating around. You need to redesign your
application so that this is not the case.
if the code is inside the class pointed to then it should not be possible to call it in any other
way except through the pointer stored in the vector, so you don't need to check it for NULL
This is how it should be done.
ANy other way and you have
Code that is very hard to maintain and very error prone.
Can you explain why and how you are possibly calling a function in a class the instantiation of
has just been deleted from memory.
|
|
|
|
|
Thanks.
To clarify, if I understand your ending question , I needed to see in some other function whether any windows of the sort I pushed onto the vector exist.
REgarding the prog structure, I later realized that I didnt need a global vector (dangerous anyways) since I was always able to get the object that had the window as a member, so I could iterate trough the parent windows and get t o the member windows (which I had been poushing back) through the pointers to the parents.
MAny thanks for the help!
ns
|
|
|
|
|
Hi,
I am using regex to lookup some data in a string. I do not have any prior experience using regular expression and would appreciate some help on the subject.
Here is my expression:
[^\.]\[(\s*\w+\s*)+\]
I am using it to match a string with square brackets where the first character is not a dot. So in the following string "[TableName].[FieldName]" I will only want to find "[TableName]". In "[Table1] INNER JOIN [Table2] ON [Table1].[Field1] = [Table2].[Field2]" I expect to find "[Table1]" and "[Table2]".
This seems to work in some cases, but miserably fails in others. For example, in the following string "[Table1], [Table2]", it only finds "[Table2]" since it expects to see some characted before the bracket.
Basically I want it to look at the first character before "[TableName]" only if there is one. I guess I need to make [^\.] optional but I was not able to find this option with regex syntax. Please help.
Thank you.
|
|
|
|
|
The part in square brackets can be matched with
\[\s*\w+\s*\]
To say "not following a dot" you need to use a look-behind assertion. With your expression, the non-dot character gets matched as part of the regex, which isn't what you're trying to get.
(?<!\.)\[\s*\w+\s*\]
This is perl 5 syntax, not sure what environment you're using.
--Mike--
Ericahist [updated Oct 26] | CP SearchBar v2.0.2 | Homepage | RightClick-Encrypt | 1ClickPicGrabber
|
|
|
|
|
Michael,
Thank you for you reply. I am using Perl5 syntax and I was using this tool to test the expression: http://jakarta.apache.org/oro/demo.html. I guess, they just did not implement it there, because I kept getting an error (Malformed Regular Expression: Sequence (?<...) not recognized. )
Do you know any other tool I can use to test the expression.
Thanks.
|
|
|
|
|
|
Im trying to allocate a string in C(non-MFC) of a specified length.
I get an error message:
illegal constant expression
when trying to create buffer.
void parse(const int size)
{
char buffer[size];
...
}
This seems very simple, but I have tried everything. How do I go about doing this?
Kevin Shaffer
kshaff03@msn.com
|
|
|
|
|
char *buffer = (char *) malloc(size);
Five birds are sitting on a fence.
Three of them decide to fly off.
How many are left?
|
|
|
|
|
The size of the array has to be known at compile time, so it can't be a variable. You need to use malloc or calloc instead.
--Mike--
Ericahist [updated Oct 26] | CP SearchBar v2.0.2 | Homepage | RightClick-Encrypt | 1ClickPicGrabber
There is a saying in statistics that a million monkeys pounding on typewriters would eventually create a work of Shakespeare. Thanks to the Internet, we now know that this is not true.
|
|
|
|
|
Thanks!
Kevin Shaffer
kshaff03@msn.com
|
|
|
|
|
<small>kshaff03 wrote:</small>
void parse(const int size)
{
char buffer[size];
...
}I think it valid in C99, not in C++.
Maxwell Chen
|
|
|
|
|
The problem is in your code
kshaff03 wrote:
void parse(const int size)
{
char buffer[size];
...
}
The problem lies in the way you've declared the array.
since char buffer[] is a static array meaning the compiler needs to know at compile time so it knows how much memory to allocate.
You're using a variable to size it. Where you should a constant or literal.
e.g.
#define MAX 100<br />
const unsigned int size = 50;<br />
<br />
inr main(void)<br />
{<br />
char foo[MAX];<br />
char bar[size];<br />
int fred[20];<br />
}
these 3 are all legal
your example should be coded as follows
void parse(const unsigned int &size)
{
char *buffer = new char[size];
...
delete[] buffer;
}
I'll send you an e-mail shortly
|
|
|
|
|
Hi,
I am trying to built an MFC application with an MDI Frame (CWinApp). Beside this, I have implemented a DLL in which I need to open an MFC based Dialog. Within the EXE, I just need to call DoModal and it works fine but here, the function that creates the dialog is in the DLL that is called by the main application.
It crashes in the function int CWnd::RunModalLoop(DWORD dwFlags) with an unhandled exception.
// pump message, but quit on WM_QUIT
if (!AfxGetThread()->PumpMessage())
{
AfxPostQuitMessage(0);
return -1;
}
By the way, I put Debugmultithreaded Dll in the settings of the application and the DLL.
What did I made wrong, please ????
MFC Beginer
|
|
|
|
|
I have two questions pertaining to serialization.
- I have nested classes but want save each class into seperate files how do go about doing this
and
- If want to able to open files in range of different extensions e.g *.csv, *.prn and *.txt. How is this achieved?
Again many thanks
Alton
|
|
|
|
|
1. Using MFC serialisation mechanisms, you don't. The assumption is made that a single file will contain the entire document. Indeed, use is made of a 'class dictionary' within an MFC archive file, unless you redo the serialisation from scratch.
2. There's a KB article (141921) which covers this nicely.
Steve S
|
|
|
|
|
I know to code it to return an object of ostream &, istream & or the classes that derived from them.
But how do you do it to return a reference to a CArchive?
Alton
|
|
|
|
|
If you're overloading the << and >> operators, chances are that you also have a CArchive& parameter. Simply return that.
Usually you'll have something like
<br />
friend CArchive& operator<<( CArchive& ar, myObj obj);<br />
friend CArchive& operator>>( CArchive& ar, myObj& obj);<br />
inside your class declaration for myObj.
So when you implement
<br />
CArchive& operator<<( CArchive& ar, myObj obj)<br />
{<br />
...<br />
return ar;<br />
}<br />
and similarly for the operator>> function.
Steve S
|
|
|
|
|
Steve S wrote:
friend CArchive& operator<<( CArchive& ar, myObj obj);
friend CArchive& operator>>( CArchive& ar, myObj& obj);
return ar;
I've those that inside a generic underived class but the compiler complains i.e. what the hell am I talking about.
do I:
- place the functions in the appropiate C++ and header files
- subclass them from CObject (which I don't want to do)
- or find the correct syntax (which is)
Alton
There's no problem, only solutions
John Lennon
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
