Page 7 of 7

Re: SwarmSpawn v2.0 - Final

Posted: Sat Jul 16, 2016 11:49 pm
by sektor2111
I'm curious if you have optimized code different, version ported/changed from NsTDM seems to not cause major issues - monster spawns or not without strange "console info". Of course I have failed again some logic - don't fix things which ain't broken - I was changing <distancelightning> even in local games which is not necessary because only Net code is bugged not local play. Because of this extra fix (removed right now) some map was crashing badly due to some display issues bla bla occludeframe etc. I keep code for On-Line (where it seems to work properly) and if something is improved at SwarmSpawner I'll see what do helps in speeding up precaching or... reducing that little lag at first spawn. It's probably client or both server and client.
What is sure noticed, when game starts and Bots enter the party, tickrate has a tiny deviation (accessing stuff: skin, mesh, etc.).

Re: SwarmSpawn v2.0 - Final

Posted: Mon Jul 18, 2016 12:23 am
by MrLoathsome
I did make a few more optimizations in this version.

Suppose I could add in a precache bit in the init code for any "Active" swarms to help with that tiny lag when
a class is first spawned.

Be aware this will NOT always precache every actor.

In that example code from the next Dropper release I posted it will precache 90-95% of actors
fine, but on some maps not everything will spawn at location(0,0,0). (Usually queens, titans etc...)

I am not so keen on trying to make multiple retries just for the precaching, so it will just get 1 attempt per class.
(*Edit: Maybe. Still screwing around with it some.)

Question: Would you like a logfile line kicked out for bDebugMode when the precache attempt fails?

Was hoping v2.0 would be the final, but suppose I can add that and make v2.1 the final......

**Edit2: Now have a way to always precache anything. This bit will be included in v2.1 FINAL!

Re: SwarmSpawn v2.0 - Final

Posted: Tue Jul 19, 2016 6:31 am
by sektor2111
MrLoathsome wrote: Question: Would you like a logfile line kicked out for bDebugMode when the precache attempt fails?
Nope, leave log alone because that's not a problem. If you'll consider it ready I'll go for a check because right now I look at some... options:
- InventorySpot - Not my goal - I have to run there for reload and not in the middle of monsters to not forget spawn behind wall problems;
- Playerstart - Next spawned player will just telefrag monster to easy;
- Removing shadow deal because I don't have any shadow in server and shadow used by me is tracking owner anyway itself adjusting size automated;
- converting to timer-free version state type based. It's what I did with majority of these Timer() things and I could figure good things (especially at MH where even regenerator is not using timer any longer any where I intend to get rid of that primitive pawnlist method but that's a different story).
Given said affinity for Monster-Games and cute options by SS aka SwarmSpawner I have done some other tweak in my DM types - simply removing Monsters from map added as decorations in unreachable locations. My Bots are trying to hunt pawns ordered and then... will loop into a spot trying to reach at them - I don't know why we do have paths ( gotta do some checks FastTrace related if doesn't have false positives ) for such pawns behind walls - bHunting bug ??? Heck, I will delete in first second all ScriptedPawns which doesn't have a trace with a nearby node - mainly cannot reach there - so I allow spawner to bring monsters only in proper locations...

Re: SwarmSpawn v2.0 - Final

Posted: Wed Jul 20, 2016 1:41 am
by MrLoathsome
Well that question is now moot since it seems to be precaching all actors with no problems now.

Have it running on my servers atm. Will probably test for a few days before I actually post the v2.1 final final.
So far I haven't noticed any issues the currrent method of building the spawnpoint list, but that doesn't mean
there aren't any... Can't recall ever spawnkilling a monster with this, but no doubt it could happen under the right

Sounds like you should have no problems adding to your custom version any minor changes I make from the v2.0 Final (NOT) version
that is currently in post 1.

Re: SwarmSpawn v2.0 - Final(Not)

Posted: Fri Aug 12, 2016 4:04 am
by MrLoathsome

Has anybody been testing the latest, apparently non-final version, of this?
Was getting some really good feedback on this briefly, then it just stopped
after the last update..... :noidea

The few days estimate I mentioned for the next update has been postponed a bit.
It will be worth it I think. No major changes to current config files will be needed.
A few performance optimizations and misc code fixes will be included.
And possibly a new optional rotating pickup mode. (Random on all 3 axis's.)

Hope to update this shortly after I get Dropper v2.0 done.

Visit my AKCoop or EXU servers for a preview of Dropper v2.0.

Re: SwarmSpawn v2.0 - Final(Not)

Posted: Sat Aug 13, 2016 9:41 pm
by papercoffee
This man deserves a tester!!!
I would be glad to help, but I can't play UT ...Reallife issues...

Re: SwarmSpawn v2.0 - Final(Not)

Posted: Sun Aug 14, 2016 7:07 pm
by sektor2111
If code is considered done I'm gonna do some conversion from timer code to state code first, and then testing it. As I could see in some last build so far, mod uses things which I have removed because I don't like Monsters behind walls which are not a threat. Mr. Flakson was missing in action in some sessions and I don't need to repeat history. Here I think I'm gonna setup a party to spawn monsters in locations where paths to players are available. I just need to think at strategy - perhaps being a nasty loaded code I'm gonna spread it into a sort of background process. Aside I'm thinking at some easy debug (sort maps purpose) broadcasting ON-Screen if options are well rated - more than 32 spawn locations confirmed, this means work and I'll see how much time I have...

Or I'll lose patience and I'll quit, or some reason will drop me away from everything, heck knows the future...

Re: SwarmSpawn v2.0 - Final(Not)

Posted: Wed Aug 17, 2016 7:39 am
by MrLoathsome
This is good. No reports of horrible crashes or lockups or log file spam horror. :P
No complaints that any of the features aren't working right.

Only report is from Sector, who is considering interesting alternate ways of using it. This is also good.

Don't lose patience or jump ahead too far. The posted code is NOT considered done.
There will be an update very soon.

Think I finished work on Dropper v2.0 tonight.
Improvements to spawning functions/pre-cache etc I made there will
be in the next SwarmSpawn release.

Re: SwarmSpawn v2.0 - Final(Not)

Posted: Wed Aug 17, 2016 5:16 pm
by sektor2111
MrLoathsome wrote:Don't lose patience or jump ahead too far.
It is about me when I start something which takes too much time, I simply quit for another simple reason: If people don't like what you do you should not spent 2 years working at something without users, that's why I'm doing small toys, I work less time/day at whatever and only what I'm doing for me is an intensive job. Probably I will do here some personal completions and not wasting 6 months for it. If less people are loving it, the work will be pointless.
Now I went a bit busy with som DM maps, BotyMan3 seems to track full paths (including through JumpSpots) due to a mutator tool firing the power of XC_Engine19 in tweaking stuff. And then, a bunch of fixes are going to take place - including fixing all Boulder and IntroDude screwed for nothing in Levels. So it's not a rush for Swarmer - take you time relaxed because I'm preparing my soup which also is requiring... Tests... More Tests... Including trying to recall what else I could see in my old logs (unsaved in that time - people were saying that are not fixable - but now they are... in big parts).
So, I'll keep an eye on this, and inspecting more maps (it helps when you do something and you must find if it will break some stuff or not).