Review Request 119564: Add define to re-enable Qt functionality we depend on.

[Alex Merry]

> On Aug. 2, 2014, 9:04 a.m., Alex Merry wrote:
> > I would rather change the code. The Qt behaviour was changed for a reason, to prevent accidental use of dangerous behaviour, and I'm not too keen on undoing that move for all software that uses KDECompilerSettings.
> 
> Sune Vuorela wrote:
>     agreed.
> 
> Kevin Funk wrote:
>     Yep. I'm also *strongly* in favor of adjusting the code instead of enabling the define.
>     
>     In fact, I thought I've fixed all of KF5. (It isn't?). 
>     There are some compile errors in code *using* KF5, which I'm trying to port ASAP.
> 
> Axel Rasmussen wrote:
>     Would we prefer just using `static_cast` like QExplicitlySharedDataPointer used to do, or shall we use `dynamic_cast` and check for `NULL` results to be extra safe? I am willing to write up a patch that fixes this issue in KService over the next day or two.

I'd expect most of the cases to be suitable for `static_cast` (if you can guarantee what it will be from the context). If in doubt, state your uncertainty in the review request description so that reviewers know to take extra care.


- Alex


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://git.reviewboard.kde.org/r/119564/#review63665
-----------------------------------------------------------


On Aug. 1, 2014, 4:06 p.m., Axel Rasmussen wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://git.reviewboard.kde.org/r/119564/
> -----------------------------------------------------------
> 
> (Updated Aug. 1, 2014, 4:06 p.m.)
> 
> 
> Review request for Build System, Extra Cmake Modules and KDE Frameworks.
> 
> 
> Bugs: 337472
>     http://bugs.kde.org/show_bug.cgi?id=337472
> 
> 
> Repository: extra-cmake-modules
> 
> 
> Description
> -------
> 
> Upstream Qt commit e112c2e altered the way QExplicitlySharedDataPointer behaves by default, such that it no longer uses a `static_cast` to cast compatible pointers. A nontrivial amount of KDE code (I encountered the build error in KService, although there are other users) depends on this functionality, so this commit adds a define to the build system which re-enables the code in Qt.
> 
> 
> Diffs
> -----
> 
>   kde-modules/KDECompilerSettings.cmake fdc930e 
> 
> Diff: https://git.reviewboard.kde.org/r/119564/diff/
> 
> 
> Testing
> -------
> 
> I applied this patch, attempted to build KService, and the build succeeded, rather than exhibiting the compile errors found in the log provided on the bug report.
> 
> 
> Thanks,
> 
> Axel Rasmussen
> 
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/kde-frameworks-devel/attachments/20140804/7f7abd5a/attachment.html>