2:38:43simplegausshi. i think i may have found an optimization bug, but i'm confused how it can even occur: https://pastebin.com/SjBV06L2 .
2:40:06simplegaussbasically if a macro expands to a function and i use it directly in a funcall, a separate lambda is compiled, but if i do the macroexpansion myself or assign the result of the macro to a function variable first (still within a funcall mind you) the code of the macro gets directly inlined (in disassembly)
10:54:26scymtymstassats: re simplegauss' question: adding a clause for macroexpand-and-retry to ENSURE-SOURCE-FUN-FORM solves the particular issue (the %FUNCALL translator has something similar with a comment to that effect). does that make sense?