Search
12:06:21
Shinmera
Fixed the Trial problem where dymanic insertions (compile-into-pass) would not properly reinstate the order.
12:06:41
Shinmera
As a side effect there's now the functions ENTER* and LEAVE* that should be used for dynamic insertions.
12:07:07
Shinmera
ENTER and LEAVE from before should only be used for things that don't require rendering, or during a static load block.
12:18:25
selwyn
what do you mean by 'dynamic insertion'
12:18:34
Shinmera
when an entity is added dynamically
12:18:44
Shinmera
while the game is running
12:18:59
Shinmera
as opposed to statically during a dedicated load zone
12:20:36
selwyn
look forward to using the new functions
12:21:16
Shinmera
I've also been thinking about ripping out Flare and replacing the indexed-set thingy with something better.
12:21:25
Shinmera
Not quite sure yet how that would work, though.
12:26:02
Shinmera
Maybe something like a flat vector with a deletion counter and compaction only once the counter gets too high.
12:26:25
Shinmera
or splitting into chunks of limited size so the maximum work per deletion is always bounded.
12:26:51
Shinmera
Could probably also get away with an unsorted table iteration for a lot of things.\
12:31:00
selwyn
why would you make such a change? for speed?
12:38:40
Shinmera
Well, 1) flare is barely used so it's a whole system and complexity that's kinda shitty to have around
12:38:56
Shinmera
2) yes, speed. The indexed set uses a doubly-linked list, which is not great for what you do most with it: iteration