Archived tweet #10357 from the @leastfixedpoint Twitter Archive

@laurencetratt Yeah totally! It's the basis for the claim at the top of p22 of doi.org/10.1007/978-3-…: "When the inline cache was enabled the performance [of a simple generic function] was better than twice that of [a switch-based implementation in] optimised C." It's a nifty technique.