[self-follow-up] At 2023-09-01T12:47:58-0500, G. Branden Robinson wrote: > > + if (curr_back_idx != back_idx) { > > + putp(tparm(set_a_background, back_idx, 0, 0, 0, 0, 0, 0, 0, 0)); > > + curr_back_idx = back_idx; > > What were people thinking? It's SO MUCH CHEAPER to pass 9 longs on > the stack than a pointer to a struct
[or array] > (which a good compiler will find a register for anyway). [but probably not on 32-bit x86] In reviewing my message I experienced a sudden horror that I may have been too subtle here. My remark was intended as the deepest possible, so-black-that-Anish-Kapoor-has-already-sent-me-a-cease-and-desist-letter sarcasm. Regards, Branden
signature.asc
Description: PGP signature