This reverts commit b87d8e5ede060e10b5927b8721933c05d1295c0d.
This didn't completeley fix the issue, also this casues merging conflicts
with some other commits so i'm reverting this.
done by using this regex replace:
'(std::bind[\ \t\n\r]*\([\ \t\n\r]*[^\,]*[\ \t\n\r]*,[\
\t\n\r]*)_([1-9])' to '$1std::placeholders::_$2' (msvc syntax)
Using boost::placeholder with std::bind seeed to cause compilation
erros with some libc++ version used by dafault on some OSX versions.
There seems to possibly be an internal issue with the slider (and more generally, the
scrollbar implementation form which it inherits) where is min/maximum_positioner_length
equal an even number, sliders with an even number of options will reach max value.
Setting it to what, in this case is 1 below the actual width of the slider image (15)
means both even and odd-numbered sliders will reach max value, however the edge is cut off.
Attempting to set the right_offset to 1 and using the above off value for the positioned
width means only sliders with an odd number of options reach max value.
The magic numbers in this commit seem to give the widget enough space to reach the end of
the slider and max values regardless of the number of options, but the internal widget dimension
calculation should perhaps be investigated in the future.
This fixes a crash that would be caused by memory incorrectly being freed
when removed from a list. Relying on splice stops that from happening.
This also fixes an exit condition when deleting handlers as depth could
sometimes not be decremented.
Finally, checks are made to made sure that an sdl_handler is either
exclusively in the global context or a different context, ensuring
that they cannot be in both.
This caused an error message and recruiting to be abandoned when the
recruit rushers Micro AI was used for two different sides in the same
scenario. It was only presents since the AI refactoring done during
1.13.4+dev and did not affect any release.
Ability to emulate floating point calculations makes sense only if we
intend to support platforms without native floating point support. Such
platforms haven't been relevant since mid-1990s or so. Besides, supporting
such platforms would require converting the entire codebase to use the
tfloat class instead of native float and double types, which would be
outright Herculean effort.
As a bonus, killing floating_point_emulation.cpp also eliminates two uses
of boost::enable_if_c.
I also removed the nonexistent file SDL_SavePNG/savepng.hpp from the
Visual Studio project, and added #include <string> to
formula/debugger_fwd.hpp because the build failed otherwise
(maybe compilation order changed when I removed
floating_point_emulation.hpp?).
This removes all the custom reference counting code from the
formula variant class, and also removes the WFL refcount() function.
The reference counting base class is also removed, since it is no longer used.
This constitutes drop-in replacements for:
* boost::shared_ptr
* boost::scoped_ptr
* boost::weak_ptr
* boost::enable_shared_from_this
* boost::static_pointer_cast
* boost::dynamic_pointer_cast
This excludes boost::intrusive_ptr, except for stray includes. Refactoring that is more complicated.
