<ELLIOTTCABLE>
devyn: on a single test-case, nonetheless!
<ELLIOTTCABLE>
re: discarding ordering,
<ELLIOTTCABLE>
the more I think about it, the more I think it's okay.
<ELLIOTTCABLE>
I'm still not adding it to the spec yet, for obvious reasons, but I think your approach is a fine middle ground, for right now.
<devyn>
ELLIOTTCABLE: the caching stuff I added reduced the time by like 75% lol, and I want to prove that parallelism is indeed possible. I don't mind if I have to redo stuff since I've been keeping optimization stuff relatively isolated - for example, caches are in a single Cache structure on the reactor
<purr>
lol
* ELLIOTTCABLE
grins
<devyn>
and I've been doing API cleanup stuff too
<devyn>
ObjectRef::new(box Thing::new()) is now Thing::empty(), etc.
<ELLIOTTCABLE>
man I want to Paws,
<ELLIOTTCABLE>
but I'm so distracted by Dwarf Fortress now :P
<devyn>
ELLIOTTCABLE: oh, and I also got rid of the React()/Yield return value stuff, so now there is only reactor.stage() and t hat's it
<ELLIOTTCABLE>
stupid game is addictive
<devyn>
and reactors have split into SerialReactor and ParallelReactor,
<devyn>
and SerialReactor just uses a queue like Paws.js
<devyn>
and is super fast
<devyn>
examples/04 executes in 4 ms
<devyn>
ELLIOTTCABLE: how much work have you done on Rulebook, btw? I'm thinking it would be cool to integrate fs watchers and have a nice GUI either via node-webkit or something in the browser over SSE
<ELLIOTTCABLE>
cli gui is more attractive to me than a local webpage
<devyn>
mmmkay
<ELLIOTTCABLE>
but no work. I haven't been worrying about it. Would rather get rulebooks written, first.
<ELLIOTTCABLE>
happy to stare at TAP output until then :P
<devyn>
heh
<devyn>
oh, also, specifications can't have comments currently, because void[] is implementation.
<ELLIOTTCABLE>
semicolons *are* void
<ELLIOTTCABLE>
completely identical
<devyn>
ELLIOTTCABLE: not completely identical. there is no equivalent to: implementation void[] "this is a comment"
<devyn>
"this is a comment"; would send that label to locals