boost/serialization/detail/get_data.hpp leaks STD macro

[Jonathan Jones <jonathan.jones@…>]

The header boost/serialization/detail/get_data.hpp defines the macro STD for local use in the header. However, it does not un-define the macro at the end of the header, causing a commonly used name to be leaked as a macro for anything included after this header.

[Robert Ramey]

OK - fixed on my machine - will eventually arrive at release

[dave.lowell@…]

The definition of STD in get_data.hpp is unconditional. What it's defined to depends on whether STLport is being used, but it's always defined to something.

However, the undef that's been added is conditional - STD is only undefined when STLport is being used. If it's not used, then it looks like the definition still leaks.

[dave.lowell@…]

patched version of get_data.hpp with unconditional undef

[Robert Ramey]

Sorry I'm not seeing this is the current code which looks like:

#if defined(__SGI_STL_PORT) || defined(_STLPORT_VERSION)
#define STD _STLP_STD
#else
#define STD std
#endif
...
#if defined(__SGI_STL_PORT) || defined(_STLPORT_VERSION)
#undef STD
#endif

[jonathan.jones@…]

Replying to ramey:

Sorry I'm not seeing this is the current code which looks like:

#if defined(__SGI_STL_PORT) || defined(_STLPORT_VERSION)
#define STD _STLP_STD
#else
#define STD std
#endif
...
#if defined(__SGI_STL_PORT) || defined(_STLPORT_VERSION)
#undef STD
#endif

What happens if neither SGI_STL_PORT nor _STLPORT_VERSION are defined?

The first block #defines STD to "std". The second block does not #undefine STD, hence the macro definition for STD leaks.

[Robert Ramey]

OK - I don't know what I was thinking.

Fixed again now