frontend: Use qobject_cast when casting QObjects

[cg2121]

Description

When casting between QObjects, qobject_cast should always be used. This increases performance as there is no RTTI (Run Time Type Information) with qobject_cast, like there is with dynamic_cast.

Using reinterpret_cast is bad practice, as you should use it only in very specific cases.

Motivation and Context

Better code

How Has This Been Tested?

compiled and ran OBS

Types of changes

• Bug fix (non-breaking change which fixes an issue) -->
• Performance enhancement (non-breaking change which improves efficiency)

Checklist:

• My code has been run through clang-format.
• I have read the contributing document.
• My code is not on the master branch.
• The code has been tested.
• All commit messages are properly formatted and commits squashed where appropriate.
• I have included updates to all appropriate documentation.

[cg2121]

Converting to draft, as #11785 needs to be merged first.

[cg2121]

This PR now builds correctly

[PatTheMav]

Overall I'm fine with this change, though it obviously touches a whole lot casts in the code that touch a whole lot of aspects of our frontend code, so this might be too late for the next version but should possibly merged after the next version is stabilised to give it some time to exist on the master branch.

@RytoEX WDYT?

[RytoEX]

Overall I'm fine with this change, though it obviously touches a whole lot casts in the code that touch a whole lot of aspects of our frontend code, so this might be too late for the next version but should possibly merged after the next version is stabilised to give it some time to exist on the master branch.

@RytoEX WDYT?

I think this should be merged for 32.1, merge timing TBD. It does have a merge conflict though.

[Warchamp7]

I've rebased this PR

[RytoEX]

This requires a rebase.

[Warchamp7]

This requires a rebase.

Rebased and conflicts resolved, however I've also added a to-be-squashed commit with some additional casts that were either introduced in the rebase, or missed originally.

[sebastian-s-beckmann]

I haven't gone through all the files (only about one third) but the idea should be clear.
There are still a lot of qobject_casts that should be static_casts. qobject_cast is only needed where it isn't clear (or assumed) that the cast succeeds. Especially in cases where the pointer is dereferenced immediately after the cast, qobject_cast doesn't make sense. Also in some cases, instead of casting from the main window it should just be OBSBasic::Get().
Also commit authorship is still changed.

[Warchamp7]

I haven't gone through all the files (only about one third) but the idea should be clear. There are still a lot of qobject_casts that should be static_casts. qobject_cast is only needed where it isn't clear (or assumed) that the cast succeeds. Especially in cases where the pointer is dereferenced immediately after the cast, qobject_cast doesn't make sense. Also in some cases, instead of casting from the main window it should just be OBSBasic::Get(). Also commit authorship is still changed.

Commit authorship is intentional since at this point it contains changes from both of us.

Will address all the feedback.

[sebastian-s-beckmann]

Seems correct

[PatTheMav]

Even though it might seem superfluous to use qobject_cast or dynamic_cast without checking the result for nullptr after, even a missing nullptr check will create deterministic crash points (mainly when that pointer is dereferenced).

But as the crash by UpdateContextBar demonstrates, static_cast will produce a valid pointer and thus dereferencing will succeed and if both types happen to have identical member or method names (like GetSource) code might potentially succeed and produce corrupted data or crash at arbitrary points later.

Otherwise I agree that where no checks and balances are used (and code is pretty blasé about safety so far anyway) using static_cast directly seems more "honest".