Previously ttree_view_node used toggle_->get_value() to know whether it is
folded or not.
This makes ttree_view_node more robust against cases where people change
the value of toggle_->get_value() manually using c++ or lua code.
This also allows us to have nodes that can only be folded via code and
have no toggle panel to expand.
Until now, [damage] (and possibly other) specials which relied on filters involving the opponent would display
in the sidebar as if the augmentation were always active (though when attacking, it worked correctly).
This is now fixed - such specials never show the augmentation in the sidebar now.
In addition, some minor code cleanup (initializing the weapon special adjacency filters outside the loop).
Slightly modified versions for wall-chasm transitions still need to be added.
Old definition for earthy hewn cave walls set as hidden but kept for compatibility; gets drawn exactly the same as the normal walls now.
In most cases, the widget size is the size of the text and looks no different
with this change. However, in cases where the label widget area is larger than
the text (such as in Unit Create), it resulted in text being rendered in the
middle of the available space, which looked weird.
For this, I re-created the earthy versions from the normal ones, so their color has very slightly changed. The difference should be subtle enough to not matter.
t_string's implicit conversion to const std::string& returns a reference
to a(n aggregate class) field that gets destroyed as soon as the
t_string is, so binding this to a const reference does not extend the
referenced object's lifetime, resulting in a dangling reference and UB.
For some unknown reason host lua script starts a sp game after mp game is finished. This commit attempts to fix thiy by giving the lua plugins a second chance to call context.exit()
The storyscreen would trigger a full redraw-cycle partially through
displaying bigmap. Due to the implementation of the storyscreen, this
would cause it to have the background overdrawn and never updated. The
fix implemented here is to remove layered-drawing functionality from
the storyscreen and trigger a full redraw manually at the end of the
story screen functionality.
The travis mp test failed. A possible cause could be that the [init_side] was not sended to the server if the scenario was ended because of a [endlevel] during a new side turn event. This commit attempts to fix that case.
