Return to “Announcements”

Post

Re: The State of Limit Theory Development, 2017 Edition!

#181
Dinosawer wrote:
JoshParnell wrote: Considering LuaJIT + multithreading is basically not happening (based on today's experiences), it seems an appropriate thing to explore :V
:cry:
*shakes fist at scripting interpreters*
*shakes fist and makes several other gestures at garbage collection*

Seriously. I hate GC. It's terrible. It ruins everything. It's slow. It's complicated. It tanks compared to refcounting. It doesn't take a genius to use weakptrs where necessary. WHO THOUGHT GC WAS A GOOD IDEA :evil:

It could all work Dino, it could all work. I could be running parallel logic right now in 16 threads. I could be happily shouting from rooftops that FPLT IS SOLVE because that extra bump from threading would get me there. But I'm not. All because of the GC.

:cry:

But here here, the night is still young, there are still many more hours for me to be defeated by this ugly beast.
“Whether you think you can, or you think you can't--you're right.” ~ Henry Ford
Post

Re: The State of Limit Theory Development, 2017 Edition!

#182
JoshParnell wrote:
Dinosawer wrote:
JoshParnell wrote: Considering LuaJIT + multithreading is basically not happening (based on today's experiences), it seems an appropriate thing to explore :V
:cry:
*shakes fist at scripting interpreters*
*shakes fist and makes several other gestures at garbage collection*

Seriously. I hate GC. It's terrible. It ruins everything. It's slow. It's complicated. It tanks compared to refcounting. It doesn't take a genius to use weakptrs where necessary. WHO THOUGHT GC WAS A GOOD IDEA :evil:

It could all work Dino, it could all work. I could be running parallel logic right now in 16 threads. I could be happily shouting from rooftops that FPLT IS SOLVE because that extra bump from threading would get me there. But I'm not. All because of the GC.

:cry:

But here here, the night is still young, there are still many more hours for me to be defeated by this ugly beast.

Oh, oh yeah. You would likely be best just going to pure C if you want to avoid the GC...

Code: Select all

<+BMRX> Silver Invokes Lewdly Verbose Experiences Readily With Absurd Rectal Expeditions
Post

Re: The State of Limit Theory Development, 2017 Edition!

#186
Dinosawer wrote:Just throwing around ideas here I got from 5 minutes of google (so indulge my possible stupidity :ghost: ), but manual memory management using LuaJIT's FFI and malloc can't help?
Sure, any memory that I manage is easy enough to get at with threads. But then I'm back to writing game objects in (essentially) C, and the usefulness of even using LJ in the first place plummets.

That being said, I have just had the first successful test of threading in LJ using some very evil but thus-far-working techniques to get Lua to share memory that it doesn't want to share. I have to be very careful here because the whole thing hinges on a hack :ghost: But it may.....it may just work....(send your prayers) :monkey:
Jelmer wrote:Glad you're back, Josh. Take it easy from now on. I can fully understand why you didn't want to come back 'empty handed'. Every day not coming back, makes coming back harder.
Well done for pushing through that barrier!
Thanks man...it does indeed. No more disappearing...it's too stressful! :shifty:
“Whether you think you can, or you think you can't--you're right.” ~ Henry Ford
Post

Re: The State of Limit Theory Development, 2017 Edition!

#188
Talvieno wrote:
JoshParnell wrote:I have to be very careful here because the whole thing hinges on a hack
And now you're coding like me! :D Told you guys I wasn't crazy. :ghost:
:V
Lua Reference Manual wrote:

Code: Select all

void lua_xmove (lua_State *from, lua_State *to, int n);
Exchange values between different threads of the same global state.

Josh's Addendum: Meh. Exchange between all the states! Just be careful not to shoot your eye out.
“Whether you think you can, or you think you can't--you're right.” ~ Henry Ford
Post

Re: The State of Limit Theory Development, 2017 Edition!

#190
I find this report as satisfactory. as Grand Admiral Thrawn would have done, conquer the New Republic with a few ISDs and some clever tactics.
See the problem as an abstraction, and solve the abstraction, then apply it into the reality of the problem. sadly Computers don't usually produce art on their own, or else we could study it and exploit it (though i have an extreemly limited knowledge of the internal workings of computers)

P.S. Welcome back to the fold
There is no peace, only passion
Post

Re: The State of Limit Theory Development, 2017 Edition!

#194
Welcome back Josh! Good to hear you're opening comms again!
I haven't posted much, but I have been lurking on these forums for a long time - anticipating your game with much ...erm ... anticipation.

Anyway, I saw your posts about the garbage collector and threads etc. Recently I've been exploring new languages that are out and about and found http://rust-lang.org/.
Apparently it's memory safe, no garbage collector and designed with threading in mind from the ground up.
So far it seems people have good things to say about it. I noted a few projects mention that they gained a considerable amount of performance from it.

So I thought maybe it makes sense to use Rust to compile a set of smaller DLLs/libraries that hook into your C code?
Downside of Rust is that it's compiled ahead-of-time so not a scripting language either and compile times are a bit slow.
However, the Mozilla mob is working hard to improve things.

I dunno, maybe this helps maybe not. :)
Good luck with the game!

Online Now

Users browsing this forum: No registered users and 1 guest

cron