Review Request: Fix memory leak in KJSEmbed when connecting signals and slots through the QObject binding

[Commit Hook]

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
http://git.reviewboard.kde.org/r/107324/#review22087
-----------------------------------------------------------


This review has been submitted with commit 533b96b6a3e20d458177445c0ac5102b045047a1 by Daniel Calviño Sánchez to branch KDE/4.9.

- Commit Hook


On Nov. 15, 2012, 10:42 a.m., Daniel Calviño Sánchez wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> http://git.reviewboard.kde.org/r/107324/
> -----------------------------------------------------------
> 
> (Updated Nov. 15, 2012, 10:42 a.m.)
> 
> 
> Review request for kdelibs.
> 
> 
> Description
> -------
> 
> A char* is used to store the signal name, and another char* is used to store the slot name. Each name is a deep copy of the data of a QByteArray, made with qstrdup. Thus, the copies must be deleted using delete[].
> 
> Before, the function callConnect returned without deleting the signal and slot names. Now, the return value is got, the strings are deleted, and then the return value is returned. The strings are needed to get the return value, so they can not be deleted before getting it.
> 
> 
> Diffs
> -----
> 
>   kjsembed/kjsembed/qobject_binding.cpp afc1989 
> 
> Diff: http://git.reviewboard.kde.org/r/107324/diff/
> 
> 
> Testing
> -------
> 
> Memory test for class ScriptingTest in KTutorial project (that is where I discovered the leak, as it uses KJSEmbed through Kross), and unit test for ScriptingTest (to ensure that everything still works as expected after the changes).
> 
> 
> Thanks,
> 
> Daniel Calviño Sánchez
> 
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/kde-core-devel/attachments/20121116/43827946/attachment.htm>