OldCoder alternate engine tests

OldCoder
Administrator
 
Posts: 346
Joined: Mon Oct 01, 2012 14:59

OldCoder alternate engine tests

by OldCoder » Sun Sep 27, 2015 00:43


These are Minetest Forums but I thought I'd cross-post the following Freeminer post as things relevant to Minetest might come up. Plus, the screenshot is amusing.

===

I'm willing to host a Freeminer world, using the Freeminer network protocol. Some debugging and tweaks may be needed first.

I've been able to build a Debian Freeminer server and a separate client for my own Linux distro, both using the latest git sources. I'm doing some preliminary testing. I've noticed these issues so far:

1. Random-number generation may not be working correctly. See the screenshot below. That is a lot of "moo".

This is a problem for Minetest as well as Freeminer. I've talked with Celeron55 and it seems possible that a particular RNG routine in Minetest is broken. However, the specific issue illustrated below is occurring for my test world only in Freeminer.

2. The Freeminer server list seems to be down today. In the client, it just says "loading". Web access shows an empty list.

3. The initial media load seems to be slow. Is the curl feature from Minetest supported or is media handled differently in Freeminer?

4. I wasn't able to build a working client, under my Linux distro, using gcc 4.9.3. proller and I determined that there was probably a low-level C++ problem involved. However, gcc 6 did work.


Image

 

User avatar
TenPlus1
Member
 
Posts: 1874
Joined: Mon Jul 29, 2013 13:38
GitHub: tenplus1

Re: OldCoder alternate engine tests

by TenPlus1 » Sun Sep 27, 2015 08:46

The mobs:register_spawn and mobs:spawn_specific have a variable called active_object_count that should limit the number of mobs an area spawns at one time, by default mobs redo sets this to 1 so you should never see this many animals in one area.
 

OldCoder
Administrator
 
Posts: 346
Joined: Mon Oct 01, 2012 14:59

Re: OldCoder alternate engine tests

by OldCoder » Sun Sep 27, 2015 10:05


TenPlus1, thank you. Yes, active object count places a limit on the number of mobs that can spawn in a given area. It applies to a 16x16x16 mapblock, doesn't it? I looked at the core code recently, but I don't recall for sure.

I've studied all levels of the code recently in connection with work on my ManyMobs "mobpack". Independently of the active object count issue, there are almost certainly ABM and/or RNG bugs in the core for both Minetest and Freeminer.

In the case of Freeminer, proller has a rough idea of what may be broken. He's restructured things and a particular test may have been optimized out due to a compiler bug. He says that he can reproduce the problem, so the answer may be known soon.

I'd like to track down a separate problem in the Minetest core. Celeron55 agrees that what I've observed shouldn't be happening. I'll post or talk to people about the issue in question as time permits.

 

User avatar
Inocudom
Member
 
Posts: 2889
Joined: Sat Sep 29, 2012 01:14
IRC: Inocudom
In-game: Inocudom

Re: OldCoder alternate engine tests

by Inocudom » Tue Sep 29, 2015 04:30

Freeminer also has serious conflicts with LuaVoxelManip. To put it simply:
Image
Basically, some chunks never get generated at all.
 

OldCoder
Administrator
 
Posts: 346
Joined: Mon Oct 01, 2012 14:59

Re: OldCoder alternate engine tests

by OldCoder » Tue Sep 29, 2015 04:53


Inocudom, one step at a time, then.

I'll need to be sure that the FM server list is working and that a FM world basically runs before I can host one. I'll also expect some attention paid to a bug that serious unless it's rare. I'll try the program, regardless, and post the results over at the other forums.

I think it's worthwhile to support efforts like this one, at least to an extent. Useful exchanges might take place between projects.

 


Return to Minetest General

Who is online

Users browsing this forum: No registered users and 35 guests

cron