mirror of
https://github.com/alliedmodders/hl2sdk.git
synced 2025-09-20 04:26:03 +08:00
Backport swap() -> V_swap() rename to avoid C++11 ADL ambiguity errors
The swap() function provided in the MathLib header was renamed to V_swap in recent Source SDK versions (e.g. 2013) to avoid causing ambiguity problems with std::swap(). But older SDK versions (such as TF2) lack this change, as they predate it. The ambiguity between MathLib's swap() and std::swap() causes considerable problems when using newer features of C++ (such as std::unique_ptr) which internally call swap() in an unqualified manner to implement move semantics: /usr/include/c++/5.2.0/bits/unique_ptr.h:342:6: error: call of overloaded ‘swap(MyType*&, MyType*&)’ is ambiguous /usr/include/c++/5.2.0/bits/move.h:176:5: note: candidate: void std::swap(_Tp&, _Tp&) [with _Tp = MyType*] hl2sdk-tf2/public/mathlib/mathlib.h:611:18: note: candidate: void swap(T&, T&) [with T = MyType*] This patch backports the swap() -> V_swap() rename from the 2013 SDK version to the TF2 SDK version, so that the TF2 SDK can be used in conjunction with C++11 features such as std::unique_ptr without difficulty. More information on why swap() isn't called in a namespace-qualified manner by standard library functions: http://en.cppreference.com/w/cpp/language/adl#Notes
This commit is contained in:
@ -1613,7 +1613,7 @@ void CDetailObjectSystem::UnserializeDetailSprites( CUtlBuffer& buf )
|
||||
buf.Get( &m_DetailSpriteDict[i], sizeof(DetailSpriteDictLump_t) );
|
||||
int flipi = m_DetailSpriteDictFlipped.AddToTail();
|
||||
m_DetailSpriteDictFlipped[flipi] = m_DetailSpriteDict[i];
|
||||
swap( m_DetailSpriteDictFlipped[flipi].m_TexUL.x, m_DetailSpriteDictFlipped[flipi].m_TexLR.x );
|
||||
::V_swap( m_DetailSpriteDictFlipped[flipi].m_TexUL.x, m_DetailSpriteDictFlipped[flipi].m_TexLR.x );
|
||||
}
|
||||
}
|
||||
|
||||
|
Reference in New Issue
Block a user