the 0.8.2 Release, positive and negative

Hello developer,

it seems that you spend a lot of time in the new release 0.8.2. So thank you for this. I want to give you some feetback. I have used the new version now since it came out. But i also used the old release 0.8.1 to compare both. Here is what i noticed:

positiv:
- Hagar: no more bullet spread with primary fire
- Electro: direct hit with secondary fire
- damage over opponent’s head
- item and powerup timeing
 
negativ:
- Hagar: i don’t like the model, too big too much rotation
- Mortar: i don’t like the model
- Rocket Launcher: i don’t like the model
- Rocket Launcher: the projectile flight, the old one was better
- it would be nice to have the oportiunity to switch between old and new models

You probably did more changes, but these are the main points i realised in the client version. An now some words to the server:

dedicated linux server:
- a 0.8.2 linux server has 10% cpu load with „top“ while a 0.8.1 server, which uses an identical server.cfg file has only 4% cpu load. Both servers used the same map. Something is wrong here.
- lots of packet lost when playing with 0.8.2 client on a 0.8.2 server
- but no packetlost playing with the same 0.8.2 client on an identical configurated 0.8.1 server
- lots of packet lost when playing with 0.8.1 client on a 0.8.2 server

Sorry to say this, but my personal conclusion is neihter to change the client to 0.8.2 nor to change the server to 0.8.2.

Finaly, don't take my post wrong, i don't want to offend someone nor do I want to degrade the value of your work. I know you do all the development in your freeetime without getting any money for it.  I hope my feetback is helpfull.

I'm too not a fan of some new weapon models, I was given a link to a pk3 containing the old mortar model and was told that there are pk3's for all previous weapons models somewhere.

Thank you for the link. Do you think you could also get *.pk3 files for Rocketlauncher and Hagar?

I haven't made my own pk3's. I guess if you download 0.8.1 you can extract the weapons models and make your own pk3. I think they are just ZIP archives.

Yep you're right, but it's not so easy because i don't know which files in which directory are needed.

The one or two things I've been bummed about this release, from hosting a server perspective:

• Buffs were enabled by default (? - they appeared without being me enabling them explicitly) which spawns random stuff like Magnet, Flight, Vampire, Speed or whatever on maps. It then appeared on CTS maps and was very capable of screwing up the leaderboards with times that should not be legitimate. It's weird because they show up on some maps, and on others they don't at all.
  
• Anonymous players can no longer participate on CTS leaderboards, and there appears to be no option to permit them.
  

Surprised some don't like the new weapon models. They are a vast improvement.

Surprised some don't like the new weapon models.  They are a vast improvement.

I do appreciate the aesthetic of the new weapons. Coming from the old models, I do find the new ones are harder to tell from afar. It's no problem for maps that I am familiar with, but in new maps, they all look similar.

That said, it seems most people like the new ones better so I really shouldn't complain.

I have only one problem with the new models, and that's the complexity of them causing a drop in rendering speed. Which is a pity.

When you force the player models to yours, you're not really making that much of a difference as, at the very least, your player model is reflected server side, so any advantage you might get from that model is given to everyone.

When you force a weapon model locally, you could make all the weapons easily give away the position of the person (eg: some visual component outside the normal area that you can see around corners, different weapon colours, etc). As this wouldn't be reflected server side (ie: it would be visible only to you), it's not what I'd call fair.

Of course, there's probably other ways to do exactly this sort of thing anyway that aren't reflected server side, so not sure if it's that much a deal.
Note: Of course, if there was simply a way to say "use old models" or "use low poly models" or something that wasn't able to be overridden, that would of course help.

Buffs are indeed enabled by default, and are available in CTS maps (though, they aren't supposed to be randomized in that mode).
Looks like it's turned off in the SMB config, and not in master: https://github.com/MarioSMB/smb-servers/...ag.cfg#L32

As for the weapon models - you may be entitled to your opinion, but this isn't a commercial company where artists are a dime a dozen... Those new amazing quality models took Morphed ~40 hours each.
They're the best thing we've seen in years (and pretty much the only art we've received in that long too), so it may be a good idea to show some respect if there's any chance of improvement or new models.

You're more than welcome to override them locally, customization is something we support here, but complaining about them being 'worse' won't get you anywhere (besides maybe Whackity Wicktown).

Personally, I wasn't a fan of the old models, partly because half of them is covered in a generic glow texture that gives them a really old look when it's supposed to modernize them... That said, I do hope Morphed may be generous enough to take another look at the colors of the new models, possibly to make them closer to the original's (i.e. a brown mortar would be nice).
Performance loss is an unfortunate part of higher quality assets, and it's made worse by the fact these models are animated. Perhaps an unanimated copy can be distributed to the lower end users.


To the OP: I haven't seen any consistent packet loss on 0.8.2 servers, and have run many myself for a long time. What are the system specifications of your server?

First of all, I didn't mean to undermine Morphed's wonderful work. As I said, I very much like how they look. I'm learning to tell them apart, which I personally find takes a bit more time than the older version (probably because of the colors?.

If most people find it OK, it probably means I can adapt too.

When you force a weapon model locally, you could make all the weapons easily give away the position of the person

Not sure I understand how, but perhaps you shouldn't elaborate.

So what? The code is open, everyone is free to create their own hacks and cheats. The only thing that's important is what is going server-side.

So what? I have worked the whole month full time on survival gamemode. Does that mean that I should be given accolades? People are free to ignore the work that doesn't suit them.

I'd be more than happy to give you accolades! Contributions are very much appreciated!

As a community, we are not only co-shaping what Xonotic becomes, but also working together to create a sustainable open-source culture here. Maintaining a welcoming atmosphere is important.

So what? I have worked the whole month full time on survival gamemode.

@Lyberta: You can have your cookie without downplaying others. Implying someone is a "sheep" isn't going to win you any favors here. Good job to you both. FWIW I greatly prefer the new models.

@Lyberta: You can have your cookie without downplaying others. Implying someone is a "sheep" isn't going to win you any favors here. Good job to you both. FWIW I greatly prefer the new models.

I'm sorry, I just have a history of wasting years for almost nothing.
Team Fortress 2 maps and mods. 2007-2012. Thousands of hours of work. Result? Valve has ruined the game and made it unplayable.
Red Eclipse maps. 2013-2016. Thousands of hours of work. Result? Got banned for disapproving the policy of devs.
My own C++ code. 2010-2017. Thousands of hours of work. Result? Nobody uses it except me.

I'm sorry you feel that way, I'm sure a lot of people in this community can empathize.  This is, however, a passionate community that generally celebrates FOSS.  With the open licenses, even if you don't feel your projects are getting love, there might be lurkers studying off your code.  One thing the team is very adamant about is providing sources.  Everyone's philanthropic code contributions help make the game and community what it is.

Can I please request we stay on the topic of the Xonotic release?  You can vent about your frustrations in a new thread, if desired.  Thanks.

To the OP: I haven't seen any consistent packet loss on 0.8.2 servers, and have run many myself for a long time. What are the system specifications of your server?

I used a dedicted server with 100Mbit, CPU:AMD Athlon II X4 605e, Quad-Core, RAM:8GB, HDD:2x1000 GB SATA II-HDD 7.200 rpm in RAID1 and Ubuntu 12.04 Server 64Bit.
Like i said before, the 0.8.1 Xonotic worked well, it was no problen to play with 10 players or more, but lots of packetlost on the same config with 0.8.2 Xonotic server only playing with bots or with another player. Not only i noticed that, all the other players said the same and packetlost could be seen in the scoreboard. I also noticed, that i had not packetlost at map boil, but extrem packetlost at map xoylent. In al cases it didn't matter what client version has been used.

Today i played here: Best of two worlds DM (88.198.253.27:26000) on map xoylent with Bots and another human players. This server uses v0.8.2 and i had a lot of packet lost there. Before we played another map also with bots and human players and there was no packetlost. So maybe its a v0.8.2  problem with some maps in addidtion with bots.

Yes, bots lag. The packet loss is happening because the server simply takes a very long time to compute bots. Try map Oilrig with 16+ bots for the ultimate lag fest.

Strange that i can run a 16 bot v0.8.1 server and play map xoylent or oilrig with nearly no packetlost or lag, but when i setup the same server as v0.8.2 it's horrible. I have the feeling the lag gets bigger the more projectils a weapon is fireing.

So what is so different in the 0.8.2 code? The bots are more intelligent and play better in v0.8.2, thats a good improvement, but not acceptable that it causes lot of packetlost. And why do have most of the maps less or nearly none packetlost and others have more?

I think the worst map is xoylent. Can something be done to avoid this lag, besides not to play xoylent or not to use bots?

I have the feeling the lag gets bigger the more projectils a weapon is fireing.

Its always been like that expectantly with vehicles. My vehicle servers would go to 50% CPU+ usage with several spider bots in an area shooting at each other min you this was with 18+ people in the server.

My question now is: will this packetlost be fixed?

There's lots of work behind the scenes by people looking into it, and I've seen code that improves the CPU usage bots have, which reduces the lag. Not sure it's ready for any sort of release though. Might be a lot more improvement to be had.

There's also some work going on in parallel about client side issues that, while it doesn't affect network traffic, do affect player experience. It's quite possible that after both are a bit more stable, there might be another release.

Note: I'm not a dev, so I can't speak for the dev's here. I am however poking around on the edges and see what they're doing.

I have not found one v0.82 Server where map Xoylent is running smooth with bots. So my question to the devs is: will this be fixed? And i'm not talkin about a fix like: ok we will delete Xoylent from the map pack. This would be a really unprofessional solution, because other maps are running similar bad with v0.82.

Would threads for darkplaces help the issue?

I run a 0.8.2 server with bots. As far as I can tell, for cases of client packet loss, it was from other network ongoings.