Conversation

Notices

1. Embed this notice
   ✧✦Catherine✦✧ (whitequark@mastodon.social)'s status on Monday, 06-May-2024 13:46:21 JST ✧✦Catherine✦✧

   wrote some CPython code today

   • Embed this notice
     val (val@oc.todon.fr)'s status on Monday, 06-May-2024 13:46:20 JST val

     in reply to

     @whitequark wakers[:] = (waker for waker in wakers if waker(*va_args))

     clacke likes this.
   • Embed this notice
     ✧✦Catherine✦✧ (whitequark@mastodon.social)'s status on Monday, 06-May-2024 13:46:21 JST ✧✦Catherine✦✧

     in reply to

     figured out how to improve it

     clacke likes this.
   • Embed this notice
     clacke (clacke@libranet.de)'s status on Monday, 06-May-2024 13:46:24 JST clacke

     in reply to

     • val
     @val @whitequark You mean it becomes equivalent to `wakers[:] = [waker for waker in wakers if waker(*va_args)]`?
   • Embed this notice
     val (val@oc.todon.fr)'s status on Monday, 06-May-2024 13:46:26 JST val

     in reply to

     @whitequark indeed, it looks like this materializes the iterable: https://github.com/python/cpython/blob/c7c9b913c01afb8d2ff4048f82155969f7ef75b1/Objects/listobject.c#L891

     sad

   • Embed this notice
     ✧✦Catherine✦✧ (whitequark@mastodon.social)'s status on Monday, 06-May-2024 13:46:27 JST ✧✦Catherine✦✧

     in reply to

     • val

     @val actually I don't think that can be the case, or it would break whenever you expand the list

   • Embed this notice
     ✧✦Catherine✦✧ (whitequark@mastodon.social)'s status on Monday, 06-May-2024 13:46:29 JST ✧✦Catherine✦✧

     in reply to

     • val

     @val oh, interesting! this does basically the same amount of allocation, right?

   • Embed this notice
     val (val@oc.todon.fr)'s status on Monday, 06-May-2024 13:46:29 JST val

     in reply to

     @whitequark probably, yes

   • Embed this notice
     Steven Reed (srtcd424@mas.to)'s status on Monday, 06-May-2024 13:46:34 JST Steven Reed

     in reply to

     @whitequark OT: my brain is apparently incapable of not reading "wakers" as a slightly differently spelled noun..

     clacke likes this.
   • Embed this notice
     ✧✦Catherine✦✧ (whitequark@mastodon.social)'s status on Monday, 06-May-2024 13:46:39 JST ✧✦Catherine✦✧

     in reply to

     • sanfierro

     @sanfierro yes

     i'm making a kind of obscure joke where i took some Python that reads similar to C code, and made it look a bit more like C code

     clacke likes this.
   • Embed this notice
     sanfierro (sanfierro@pony.social)'s status on Monday, 06-May-2024 13:46:40 JST sanfierro

     in reply to

     @whitequark
     > va_args
     I think I've seen this somewhere, perhaps C?
     Also, I think you've just shown me what wakers are, but I still don't fully understand

   • Embed this notice
     ✧✦Catherine✦✧ (whitequark@mastodon.social)'s status on Monday, 06-May-2024 13:46:54 JST ✧✦Catherine✦✧

     in reply to

     • clacke
     • val

     @clacke @val it's probably somewhat slower

   • Embed this notice
     clacke (clacke@libranet.de)'s status on Monday, 06-May-2024 13:46:54 JST clacke

     in reply to

     • val

     Yeah, there are more objects created and I imagine the generator protocol calls aren't cheap.

     Oops also it would break if the filtered list were empty. Guarding it by adding index = -1 before removes the golf benefit.

     Once again, explicit is better.

     @whitequark @val

   • Embed this notice
     clacke (clacke@libranet.de)'s status on Monday, 06-May-2024 13:46:55 JST clacke

     in reply to

     • val

     @whitequark This should have the same memory properties as the original:

     for index, waker in enumerate(waker for waker in wakers if waker(*args)): wakers[index] = waker del wakers[index + 1:]Not sure if it's more readable or if I'm just liner golfing. I'm also not thrilled with using index outside the loop, I actually wish that wouldn't be in scope.

     @val

   • Embed this notice
     ✧✦Catherine✦✧ (whitequark@mastodon.social)'s status on Monday, 06-May-2024 13:46:56 JST ✧✦Catherine✦✧

     in reply to

     • clacke
     • val

     @clacke @val basically yes, as far as i can tell from the code