freenode/#shirakumo - IRC Chatlog

14:54:08 Shinmera Figuring out how to make this more efficient is quite difficult

14:54:54 Shinmera Trying to optimise these two functions https://github.com/Shinmera/uax-9/blob/master/uax-9.lisp#L143-L192

14:54:55 Colleen github.com/Shinmera/uax-9/b... Website (HTML), Title: uax-9/uax-9.lisp at master · Shinmera/uax-9 · GitHub

14:59:40 Shinmera Profiler tells me determine-level-runs is hot, which is not surprising.

15:00:00 Shinmera https://plaster.tymoon.eu/view/1497#1497

15:00:00 Colleen plaster.tymoon.eu/view/1497... Website (XHTML), Title: - Plaster

15:00:51 Shinmera make-status-stack especially, which is just a struct allocation, but it can't be stack-allocated :(

15:03:04 Shinmera Hmm, I suppose I can take out the array allocation and DX that instead, then pass it to the struct.

15:03:28 Shinmera Could do away with the struct entirely by using the first field as the idx

17:07:45 Shinmera Well, anyway. https://shinmera.github.io/uax-9/

17:07:45 Colleen shinmera.github.io/uax-9/ Website (HTML), Title: Uax 9

18:05:41 selwyn Shinmera: do you believe that it is possible to correctly implement the cliff click hash table in luckless? without the excessive spinning you mentioned

18:05:58 Shinmera I don't see why not

18:06:30 Shinmera I probably just made some mistakes during porting. Scanning through the code and ensuring it's semantically equivalent is an arduous task.