TermBuf::Blend

Stability: stable — changes only in a major release.

How a write settles the style of each cell it lands on: given the style already there, the style being written, and the cell's position, it returns the style to place.

dim = TermBuf::Style.blend { |under, over| under.merge(over).faint }
screen.write 0, 0, "over a shadow", Style::DEFAULT, blend: dim

Style::KEEP_BACKGROUND and Style::OVER cover the two common answers, and Style.blend wraps a block that only wants the two styles.

The position is in the buffer's coordinates, not those of whatever view the write went through: a view translates the write before the buffer runs the blend, so a blend keyed on position sees where the cell actually is.

A blend is run per cell and its answer is interned, and StyleTable only grows. One returning a colour computed from the position — a gradient — therefore interns a style per cell. That is bounded by the screen for a single frame and fine; recomputed every frame from different colours it is unbounded, and an animation wanting that should draw from a fixed palette of styles instead.