Random Chitchat 2012-2016

[xoft]

I'm still having trouble with your world, even when set to Floody. Both the server and the client choke on it, each hogs one core of the CPU for itself. The client is even worse, I get like 1 frame per 10 seconds, then it freezes for another 10 seconds...

It seems that when more than a few (tens at most) blocks change in each chunk for a client, the client cannot keep up with the incoming BlockChange packets. It even happenned that I turned off the server and the client kept displaying the game for full two more minutes before realising that the connection has been long ago closed.

The server, on the other hand, chokes on the sheer amount of fluid simulating it has to do. There are several hundred blocks in *each* chunk scheduled for simulating in each wave. I did some slight performance-related cleanups, but still the server is choking. I'm running out of ideas here.

[NiLSPACE]

Maybe thats the reason that a vanilla nether is 128 blocks high

[xoft]

No, it's just that you have too many lava springs, and the very root cause is the Floody simulator - it changes too many blocks. The vanilla simulator doesn't spread the fluids so much so it's less strain.

[xoft]

I have received my 512MiB RasPi, so now I can try MCServer on the Pi again. This time I'm keeping the Pi, so I should be able to keep an eye on the Pi version, as well as on the performance there.

I'm also considering using an old laptop for a linux cross-build environment for the Pi, so that the compilation doesn't actually run on the Pi (and thus is faster ) The full compilation took about three hours and, boy, was I glad that I mounted some heatsinks prior to running it, the Pi was scorching hot!

[NiLSPACE]

i was thinking, is it not possible to upload MCServer to google play since you can now have limited worlds?

[xoft]

I have found some serious performance reserves in the blockhandlers, I'm gonna try fix those.

As for the Android version, so far only FakeTruth succeeded in compiling it, so he's the only one who can upload it

[xoft]

Wow! This is too good to be true! I must have broken something. Or were the performance reserves in the blockhandlers really that serious?

Previous version choke on the Nether world with 45% chance of lava springs, using an entire core of my quadcore for physics simulation.
What I have now works like a breeze, same world, same conditions and it's using less than 1 % of the CPU.

Speechless.

---

Well, I tried out a few things and nothing really seems broken, so I'm committing it to SVN. Have a whack at it, all of you, and tell me what I broke
Rev 1278.

[NiLSPACE]

the nether also needs nether quarts and soal sand generated. that should be easy if you made another OreNests but then for the nether but i think that mushrooms and glowstone would be a lot harder.

[NiLSPACE]

i found something about nether portals that could be usefull. http://www.minecraftforum.net/topic/3458...l-science/

[xoft]

Nice find, I knew about 3/4 of that already, but they give exact numbers and algorithms, that's always good.
Now we need to replicate that Though I don't think portals are gonna work anytime soon, there's just too many other things to do. Perhaps a plugin with simpler rules may do the work in the meantime.

---

Now I'm poking at generators, changing the function signatures a bit so that it's easier to use cBlockAreas in the generators.