> On 2009-07-13 10:55:33, Aleix Pol wrote:
> > it is not necessary to check if the object is null before deleting. code
> > like:
> >
> > if(a) delete a;
> >
> > should be just "delete a;"
> >
> > In C++ the null is 0, I think it's better to assign 0 than NULL.
This is not the problem, the problem is:
delete a; // in one class method
// In another class method:
if (a) // the result will the true even though a is deleted
do a->"something" // this can crash the application
I am more used to C than C++, I can change it to 0 if you wish. There
are a
lot of places in Kopete's source code that also uses NULL, and other places
that use 0 and other places that uses 0L. Which one is the best to use (0 or
0L)?
- Lamarque
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
http://reviewboard.kde.org/r/1004/#review1568
-----------------------------------------------------------
On 2009-07-13 13:08:03, Lamarque Souza wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> http://reviewboard.kde.org/r/1004/
> -----------------------------------------------------------
>
> (Updated 2009-07-13 13:08:03)
>
>
> Review request for Kopete.
>
>
> Summary
> -------
>
> Reinitialise deleted pointers to NULL. Throughout all Kopete source code (not
> only WLM protocol) pointers are deleted and not reinitialized, grep -r
> 'delete ' $(find \( -name "*.cpp" -o -name "*.h" \) ) | wc returns 1105
> lines, I can do a rough estimation of at least half of those lines do not
> reinitialize pointers. Althouth reinitialing pointers are not always
> necessary in some cases it does is necessary and by what I have seen there
> are such places in Kopete's source code. Since analising 1105 linhes of code
> is a really tiring and boring task I did it only for the WLM protocol for
> now. Kopete::CommandHandler::slotExecFinished even passed one deleted pointer
> to a function, which seems really wrong to me, it should delete the pointer
> after calling the function, not before. To illustrate the problem assume a is
> a class member pointer:
>
> delete a; // in one class method
>
> // In another class method:
> if (a) // the result will the true even though a is deleted
> do a->"something" // this can crash the application
>
>
> Diffs
> -----
>
> /trunk/KDE/kdenetwork/kopete/protocols/wlm/wlmchatmanager.cpp 993925
> /trunk/KDE/kdenetwork/kopete/protocols/wlm/wlmlibmsn.cpp 993925
> /trunk/KDE/kdenetwork/kopete/protocols/wlm/wlmsocket.cpp 993925
>
> Diff: http://reviewboard.kde.org/r/1004/diff
>
>
> Testing
> -------
>
>
> Thanks,
>
> Lamarque
>
>
_______________________________________________
kopete-devel mailing list
kopete-devel@kde.org
https://mail.kde.org/mailman/listinfo/kopete-devel
