.:: Bots United ::.

.:: Bots United ::. (http://forums.bots-united.com/index.php)
-   Sandbot (http://forums.bots-united.com/forumdisplay.php?f=85)
-   -   progress (http://forums.bots-united.com/showthread.php?t=10041)

RoboCop 24-03-2018 19:20

Re: progress
 
Also are you able to program and create those bot "Grapple" waypoints for like op4ctf_xendance? Because that map will make the bots useless unless they know how to use the "Batman's Rope" to grab from point island to another to capture flag - quicker before your bots say "good thing, Batman" Dadadadadada Batman!

tschumann 25-03-2018 00:00

Re: progress
 
Quote:

Originally Posted by RoboCop (Post 67029)
Well its a shame ShrikeBot isn't open source and it doesn't seem likely that SandBot will work with ShrikeBot waypoints.

Yeah, waypointing is the least fun part of it all.

Quote:

Originally Posted by RoboCop (Post 67030)
Also are you able to program and create those bot "Grapple" waypoints for like op4ctf_xendance? Because that map will make the bots useless unless they know how to use the "Batman's Rope" to grab from point island to another to capture flag - quicker before your bots say "good thing, Batman" Dadadadadada Batman!

Yeah I guess that's something I'll look at.
I think 007 Nightfire bots (most likely by botman) had the ability to use grappling hooks.

Persephone 25-03-2018 09:19

Re: progress
 
Have you ever tried jk_botti ? The bot that was created by Ghost of Evil Spy

They were capable of using almost all weapons including Grapple and even PENGUINS

They were still using Grapple to their opponents instead of Xen object though

RoboCop 26-03-2018 20:30

Re: progress
 
Quote:

Originally Posted by Persephone (Post 67034)
Have you ever tried jk_botti ? The bot that was created by Ghost of Evil Spy

They were capable of using almost all weapons including Grapple and even PENGUINS

They were still using Grapple to their opponents instead of Xen object though

Yes I'm aware that jukivili/Hullu made JK-Botti, as well know how clever and lightweight they work. Not to forget their unique AutoWaypoint and using almost all Op4 weapons even "Linux bombs", but don't work on Op4CTF maps.

It's just a matter of C++ skill and time to know how and when it will support those Grapple and that.

tschumann 05-04-2018 13:36

Re: progress
 
I haven't tried jk_botti.

I'll get to it eventually - I've played around a bit with the weapon code but I haven't really got to the bottom of it yet (I tried adding rocket launcher usage for Day of Defeat but I haven't worked enough of it out to make it work yet).

tschumann 06-04-2018 14:41

Re: progress
 
RoboCop could you post the results of the following command on your server:
strings /usr/lib/i386-linux-gnu/libstdc++.so.6 | grep CXX

RoboCop 07-04-2018 02:45

Re: progress
 
Expect a bit of a delay one of my dedicated servers died 2 weeks ago and I'm kinda struggling to catch up with the backup on the raw MySQL data files. But the Op4CTF server is unaffected as that one is remained safe...for now...

tschumann 07-04-2018 06:30

Re: progress
 
Alright so I've asked elsewhere and got what looks like a very good hint - it looks like SteamCMD (which RoboCop and nubizor are both using I believe) ships with an old version of libstdc++.so.6 which even now only provides up to CXXABI_1.3.7 - I'm no Linux export but I'm guessing that SteamCMD sets its own LD_LIBRARY_PATH which means that its own older libstdc++.so.6 gets loaded instead.
Could you try a version of Sandbot that didn't work before after removing steamcmd/linux32/libstdc++.so.6 (back it up just in case)?

RoboCop 10-04-2018 16:17

Re: progress
 
Well I've renamed that steamcmd/linux32/libstdc++.so.6 as libstdc++.so.6.bak as well loaded GDB and it did crash after 2-5 mins, but didn't gave me any backtrace logs. All it had was the "FATAL ERROR (shutting down): NUM_FOR_EDICT: bad pointer" and "FATAL ERROR (shutting down): MSG_ONE or MSG_ONE_UNRELIABLE with no target entity"

tschumann 12-04-2018 13:58

Re: progress
 
Alright great, I'm glad the libstdc++ problem is sorted.

Do you have a console dump from the lead-up to the crash?

RoboCop 14-04-2018 01:19

Re: progress
 
Cannot seem to have it as well have no backtraces as well...

tschumann 14-04-2018 05:32

Re: progress
 
Can you the game with -condebug or whatever and paste the end of qconsole.log when it crashes?

RoboCop 15-04-2018 02:39

Re: progress
 
I've just done that as well.

But, interestingly I've switched from Metamod-P38 to P37 that doesn't have the eagle1.sc and eagle2.sc required for the client to play on my Linux Op4CTF server. It appears to be more stable with Sandbot just now...Maybe then with P37 will work better, and I hope will fix those glitches in CTF mode...

tschumann 15-04-2018 08:50

Re: progress
 
Ah nice.

Anyway, once I've got some time I'll try and do some final testing and tag a release.

RoboCop 15-04-2018 15:58

Re: progress
 
Yes well my Linux server still have those unwanted eagle1.sc and eagle2.sc files but they crash if I remove them. And I guess it's up to Valve to fix those Op4CTF bugs and get them outta way for ourselves to focus on how well SB performs.

Also for Op4CTF server owners I recommend NOT adding those eagle1/2.sc files in your map .res files for yor players to auto download, as that will nerf the Desert Eagle's gunfire sounds and animation. And I don't mean making your Desert Eagles turn into those toy Nerf Foam Dart guns!

tschumann 18-04-2018 14:41

Re: progress
 
What needs those files? I've only got events/eagle.sc (which the game needs).

RoboCop 18-04-2018 22:30

Re: progress
 
Mainly for Linux Op4 its events/eagle1.sc and events/eagle2.sc. Valve forgot to fix that issue for Linux Op4 so many years ago.

tschumann 21-04-2018 09:20

Re: progress
 
0.4.2 is out now - couple of bugs already:
- Singleplayer doesn't work when running MetaMod - overlooked during testing but my attempt at fixing just yields errors (Can't find address: 0c7db6d0 ERROR: Invalid function pointer in entity!)
- Changing bot_count doesn't work when running MetaMod - for some reason the cvar_t doesn't get the updated value in either the string or value attributes and I've got no idea why.

Anyway, I'd spent 3+ hours manually verifying and fixing bugs and wanted to get a release out so I've released anyway.
MetaMod again succeeds at making things more difficult.

RoboCop 21-04-2018 17:56

Re: progress
 
1 Attachment(s)
Well it seems to work better if I remove the libstdc++.so.6 provided by SteamCMD. The bot combat skills need a few fixes even at bot_skill 1.

Also you missed my new Op4CTF waypoints that I've made for you. Nice work tho!

tschumann 22-04-2018 01:02

Re: progress
 
Quote:

Originally Posted by RoboCop (Post 67055)
Well it seems to work better if I remove the libstdc++.so.6 provided by SteamCMD. The bot combat skills need a few fixes even at bot_skill 1.

Also you missed my new Op4CTF waypoints that I've made for you. Nice work tho!

What's wrong with bot_skill?
And yeah I'll add those waypoints in a future release.

Still not sure what's on with the singleplayer stuff in MetaMod though - anyone successfully integrated this with their bot? Given that it works without MetaMod I'm thinking that the extra layer of indirection is breaking things, but I think the .dll files are all loaded in the same address space so I think it should work.
Also, my attempt at fixing it yesterday gave more sane messages but I think what I did was wrong (or at least looking at the singlep plugin, they did what I had originally).

RoboCop 22-04-2018 14:45

Re: progress
 
Nothing much its just they don't seem to combat at the best it can at bot_skill 1, unlike the other HL1 bots I've used...

The Storm 23-04-2018 01:18

Re: progress
 
E[POD]Bot is working fine with Metamod and Singleplayer. You can check there.

RoboCop 24-04-2018 22:57

Re: progress
 
I just had this crash from your latest SB 0.4.2 build:-

Code:

../common/singleprocesspipe_any.cpp (200) : Assertion Failed: m_Peeked.second == false || m_nEventCount > 0

Program received signal SIGTRAP, Trace/breakpoint trap.
[Switching to Thread 0xf1a58b40 (LWP 14229)]
0xf2e87a0a in ?? () from ./steamclient.so
(gdb) bt
#0  0xf2e87a0a in ?? () from ./steamclient.so
#1  0xf256f900 in ?? () from ./steamclient.so
#2  0xf2569c78 in ?? () from ./steamclient.so
#3  0xf255ffac in ?? () from ./steamclient.so
#4  0xf25626aa in ?? () from ./steamclient.so
#5  0xf2563536 in ?? () from ./steamclient.so
#6  0xf2e8f420 in ?? () from ./steamclient.so
#7  0xf2e8e09b in ?? () from ./steamclient.so
#8  0xf2e8e25d in ?? () from ./steamclient.so
#9  0xf2e90b63 in ?? () from ./steamclient.so
#10 0xf7e9af72 in start_thread () from /lib/i386-linux-gnu/libpthread.so.0
#11 0xf7dd143e in clone () from /lib/i386-linux-gnu/libc.so.6
(gdb)


tschumann 26-04-2018 13:44

Re: progress
 
Quote:

Originally Posted by RoboCop (Post 67057)
Nothing much its just they don't seem to combat at the best it can at bot_skill 1, unlike the other HL1 bots I've used...

Hm, coincidentally I just tidied up the bot_skill code and noticed that bot_skill 1 wasn't using the top level of accuracy available.

Quote:

Originally Posted by The Storm (Post 67058)
E[POD]Bot is working fine with Metamod and Singleplayer. You can check there.

Thanks I'll check it out, though I thought E[POD]Bot was for Counter-Strike and not for any single player game.

Quote:

Originally Posted by RoboCop (Post 67061)
I just had this crash from your latest SB 0.4.2 build:-

Code:

../common/singleprocesspipe_any.cpp (200) : Assertion Failed: m_Peeked.second == false || m_nEventCount > 0

Program received signal SIGTRAP, Trace/breakpoint trap.
[Switching to Thread 0xf1a58b40 (LWP 14229)]
0xf2e87a0a in ?? () from ./steamclient.so
(gdb) bt
#0  0xf2e87a0a in ?? () from ./steamclient.so
#1  0xf256f900 in ?? () from ./steamclient.so
#2  0xf2569c78 in ?? () from ./steamclient.so
#3  0xf255ffac in ?? () from ./steamclient.so
#4  0xf25626aa in ?? () from ./steamclient.so
#5  0xf2563536 in ?? () from ./steamclient.so
#6  0xf2e8f420 in ?? () from ./steamclient.so
#7  0xf2e8e09b in ?? () from ./steamclient.so
#8  0xf2e8e25d in ?? () from ./steamclient.so
#9  0xf2e90b63 in ?? () from ./steamclient.so
#10 0xf7e9af72 in start_thread () from /lib/i386-linux-gnu/libpthread.so.0
#11 0xf7dd143e in clone () from /lib/i386-linux-gnu/libc.so.6
(gdb)


Hm, not sure what's going on here but it looks like a crash is Steam? Do you have a debugger attached?

RoboCop 26-04-2018 15:44

Re: progress
 
Well I don't know if you can do this but is it possible for you to compile a Debug build of Sandbot in order to see the specific details and key cpp files triggering the crash?

As for EPB I did recompile that build couple years ago but didn't do any significant changes.

https://apg-clan.org/vbdownloads.php...ownloadid=1003

tschumann 28-04-2018 06:23

Re: progress
 
Quote:

Originally Posted by RoboCop (Post 67064)
Well I don't know if you can do this but is it possible for you to compile a Debug build of Sandbot in order to see the specific details and key cpp files triggering the crash?

As for EPB I did recompile that build couple years ago but didn't do any significant changes.

https://apg-clan.org/vbdownloads.php...ownloadid=1003

You can run make symbols and you'll get a debug sandbot.so (or at least it still has symbols in it).
But the stack-strace looks like a problem with Steam - when does it happen? It looks like something is spawning a process (the clone at the top) and I guess either the parent or child process is crashing for some reason.

tschumann 17-08-2018 14:20

Re: progress
 
It's been a little while but I'm still working on it.
The other day someone reported that savegames in single player are blank and I'm working on a fix for that.
I also want to get Opposing Force Capture Point mode working but haven't had a whole lot of time to take a close look at that.

Persephone 18-08-2018 14:27

Re: progress
 
I'm glad if this bot is still being worked 😄

Let me know if you have any updates

tschumann 23-09-2018 09:48

Re: progress
 
Finally sorted out my MetaMod issues - seems like using the raw cvar_t doesn't work in MetaMod for some reason, but directly querying the engine does.

tschumann 09-12-2018 08:37

Re: progress
 
It's been quiet but mainly because I've been working hard on another project (https://www.moddb.com/mods/sandpit).
I want to get Opposing Force Capture Point working and release that soon.

Persephone 18-12-2018 16:32

Re: progress
 
Sandpit? It looks like Garry's MOD :/

Don't forget about Sandbot please :D

Btw, have you read my private message to you?

tschumann 22-12-2018 00:29

Re: progress
 
The joys of working on multiple things at once...

What private message? I don't think I got one.

Persephone 22-12-2018 12:10

Re: progress
 
Yeah, but working on multiple projects at once could makes some headache though 😁

About private message, try click on your own profile

tschumann 22-12-2018 12:19

Re: progress
 
Yeah no private messages - the last one I got was in June.

Persephone 22-12-2018 12:38

Re: progress
 
Sorry, I mean Visitor Message, not private Message

Click on your name, then View Public Profile

tschumann 23-12-2018 05:53

Re: progress
 
I see it now.

tschumann 27-12-2018 08:27

Re: progress
 
Did some work on Opposing Force Capture Point support today - bit of a pain as there's no easy way to see who a capture point actually belongs to, if anyone, as far as I can see. pev->skin looks like it store the team owner but it doesn't.
Looks like I'll have to get all capture points and see whether the func_wall they activate when captured is visible or not... which kind of sucks.

Persephone 30-12-2018 21:47

Re: progress
 
Huh? What func_wall? I was often making a map for Opposing Force CTF with Valve Hammer Editor, and I never used it. I mean, what usefulness of it? All you have to do is placing an entity called "item_ctfbase" and it has default models called "civ_stand.mdl" (for Black Mesa team) or you can rename it to "mil_stand.mdl" (for Opposing Force team). Don't forget to place entity called "item_ctfflag" inside the base for each team. So you don't need func_wall at all.

Roberto 31-12-2018 01:00

Re: progress
 
It's hard to make bots do anything well other than DM but keep trying, tschumann! 'Cause most often the most fun modes to play are not deathmatch.

tschumann 31-12-2018 07:29

Re: progress
 
Quote:

Originally Posted by Persephone (Post 67258)
Huh? What func_wall? I was often making a map for Opposing Force CTF with Valve Hammer Editor, and I never used it. I mean, what usefulness of it? All you have to do is placing an entity called "item_ctfbase" and it has default models called "civ_stand.mdl" (for Black Mesa team) or you can rename it to "mil_stand.mdl" (for Opposing Force team). Don't forget to place entity called "item_ctfflag" inside the base for each team. So you don't need func_wall at all.

Opposing Force Capture Point is different to Opposing Force Capture The Flag - it has two trigger_ctfgeneric entities that trigger the visibility of func_wall entities.

Quote:

Originally Posted by Roberto (Post 67259)
It's hard to make bots do anything well other than DM but keep trying, tschumann! 'Cause most often the most fun modes to play are not deathmatch.

Agreed!


All times are GMT +2. The time now is 09:50.

Powered by vBulletin® Version 3.8.2
Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.