ZK lobby server split is a disaster? - Page 3

ZK lobby server split is a disaster?

Various things about Spring that do not fit in any of the other forums listed below, including forum rules.

Moderator: Moderators

User avatar
KingRaptor
Zero-K Developer
Posts: 838
Joined: 14 Mar 2007, 03:44

Re: ZK lobby server split is a disaster?

Post by KingRaptor » 27 Jan 2015, 02:09

BlitzTank is [Fx]Drone
0 x

gajop
Moderator
Posts: 3023
Joined: 05 Aug 2009, 20:42

Re: ZK lobby server split is a disaster?

Post by gajop » 27 Jan 2015, 07:08

That's why I think that each game having its own forums for players is a good thing. When there are few players it's usually OK as most of them are power users/devs, but it often gets toxic when it increases to large amounts (less visible in certain cases like grand strategy games that attract a different demographic, but still present).
0 x

User avatar
1v0ry_k1ng
Posts: 4656
Joined: 10 Mar 2006, 10:24

Re: ZK lobby server split is a disaster?

Post by 1v0ry_k1ng » 27 Jan 2015, 11:25

As qray said, ZK has been it's own entity for a long time, ZKL doesn't even list other games or join the spring channels unless you tell it to. I think the majority of ZK players don't even realize the server changed because nothing is different. The only difference now is that I can't check every once in a while to see if BA players are doing anything
This isn't wrong, though. By what criteria is the server split a disaster? ZK still works. Main spring server still works. Disaster therefore is?
0 x

Orfelius
Posts: 103
Joined: 17 Nov 2014, 20:57

Re: ZK lobby server split is a disaster?

Post by Orfelius » 27 Jan 2015, 11:58

I feel like some people think this is a disaster because Z-K has potential to become ever so popular and split makes the discovery of other spring games less possible.
0 x

User avatar
Jools
XTA Developer
Posts: 2804
Joined: 23 Feb 2009, 16:29

Re: ZK lobby server split is a disaster?

Post by Jools » 27 Jan 2015, 12:35

There's a lot of talk of "spring" and "games" here but I'm not sure what you guys are referring to, I mean, there's decrepit old BA which has been stuck in an 8v8dsd timeloop for about a decade now. There's a few little 'mods' like evo and tech annihilation which nobody plays and then there's ZK, which is the only properly developed "game" on this engine with any potential.
Dear Sir,

I wish to complain in the strongest possible form about the previous poster that portrays ZK players as lumberjacks. Many of my best friends are ZK players and only a few of them talk about their "potential" and their "game". The rest have quite normal heterosexual relationships with animals.

I also wish to register a complaint about the portraying of BA players as hamsters hanging out in the deserted river. I also know quite a few BA players and only 20% regularly visit the delta. The rest have quite normal relationships with the forest.

Your Faithfully,

Sir Jools (Ms.)

PS: I found this:
Image
Last edited by Jools on 27 Jan 2015, 12:37, edited 1 time in total.
0 x

User avatar
PicassoCT
Journeywar Developer & Mapper
Posts: 10223
Joined: 24 Jan 2006, 21:12

Re: ZK lobby server split is a disaster?

Post by PicassoCT » 27 Jan 2015, 12:36

I find outrageous that you removed the whole future JW_Playerbase from the server. Cant you for once jump over the Zero-Kredit fence?
0 x

User avatar
1v0ry_k1ng
Posts: 4656
Joined: 10 Mar 2006, 10:24

Re: ZK lobby server split is a disaster?

Post by 1v0ry_k1ng » 27 Jan 2015, 15:11

Image

I look at that list of maps and cant help but feel that the exodus was... necessary.

Image

Zero K: [BA has broken through to the surface where Zero K and his soldiers wait for them to be drowned] Let go! Don't you understand? It's for the good of the colony!

BA: What are you saying? We are the colony!

[Zero K is about to strike BA when Spring Devs knocks his fist aside]

Zero-K: Spring Devs, what are you doing?!

Spring Devs: Something I should have done a long time ago.

[extends his hand to the BA players playing 8v8 DSD in an endless limbo]

Spring Devs: *This* is for the good of the colony, General.

Zero-K: You useless, ungrateful maggot! I am the colony!!
0 x

raaar
Metal Factions Developer
Posts: 812
Joined: 20 Feb 2010, 12:17

Re: ZK lobby server split is a disaster?

Post by raaar » 27 Jan 2015, 15:53

Orfelius wrote:I feel like some people think this is a disaster because Z-K has potential to become ever so popular and split makes the discovery of other spring games less possible.
yep. And that was probably deliberate.

personally some good came out of it, though. I stopped playing ZK and have been more focused on developing my game.
0 x

User avatar
Forboding Angel
Evolution RTS Developer
Posts: 14585
Joined: 17 Nov 2005, 02:43

Re: ZK lobby server split is a disaster?

Post by Forboding Angel » 30 Jan 2015, 21:05

Actually, for evo in some ways it's been a net positive.

The ZK crowd was on it's way to becoming worse than the BA crowd and the AA crowd, so in some ways it's kinda nice now.

I remember when the official server was lucky to have 5 people logged in at the same time. It grew then, it will grow again.
0 x

User avatar
PicassoCT
Journeywar Developer & Mapper
Posts: 10223
Joined: 24 Jan 2006, 21:12

Re: ZK lobby server split is a disaster?

Post by PicassoCT » 31 Jan 2015, 11:19

A open source guy strands with a survival pod on a habitable planet.
What will happen?

Cytokinese
Drama
AntiSocial Versionoverpopulation
0 x

User avatar
1v0ry_k1ng
Posts: 4656
Joined: 10 Mar 2006, 10:24

Re: ZK lobby server split is a disaster?

Post by 1v0ry_k1ng » 02 Feb 2015, 12:12

The ZK crowd was on it's way to becoming worse than the BA crowd and the AA crowd, so in some ways it's kinda nice now.

Image
0 x

User avatar
AF
AI Developer
Posts: 20669
Joined: 14 Sep 2004, 11:32

Re: ZK lobby server split is a disaster?

Post by AF » 03 Feb 2015, 02:10

Wow, even I've played those maps to excess...

Though I wonder why nobody has ever made Super Delta Heck Ball Dry 2
0 x

User avatar
PicassoCT
Journeywar Developer & Mapper
Posts: 10223
Joined: 24 Jan 2006, 21:12

Re: ZK lobby server split is a disaster?

Post by PicassoCT » 03 Feb 2015, 10:53

Its quieter without the ZK-Crowd- through nightwatch is a nice sock-puppet.

With quietness there is less chance for social collissions, thus less chance to meet somebody who might be or become a good coder. Not everyone is learning about spring via checking out githubs-ass.

So all in all a loss, but at least one on good terms. All people in the end, want to fence of there own little world. Though i would really prefer it, if the devs team stayed one the engine server communication wise.
Reason. No reason. Just feelings.
Feelings that one doesent learn about the crap that bugs the other if you dont run into each other on a daily basis.
So worries and development diverge over time.

Marriage Councelor Voice: You need to have mutual acceptance and respect for the needs of the few outweighting the needs of the many, which sometimes might include a dwarf orgy in the living room.

On a serious note, this rocked the boat. We should have taken the performance demands of zero-k more serious. And it is actually taken serious right now.
And its a good thing that devs have to care for there own playerserverservice.

Finally to all thos moved out devs- could you form a advertisement circle? So for example - that if somebody doesent like your game - he can click on a banner advertising evo or s44 or BAR and get forwarded to it?
0 x

gajop
Moderator
Posts: 3023
Joined: 05 Aug 2009, 20:42

Re: ZK lobby server split is a disaster?

Post by gajop » 03 Feb 2015, 11:50

PicassoCT wrote:We should have taken the performance demands of zero-k more serious. And it is actually taken serious right now.
You make it sound like there were no concerns on our side regarding performance and that's the reason Licho forked it.
It's really annoying when people keep making up history.
We responded to his concerns immediately, and it's not like this wasn't considered before, or you think I just came up with this on the spot? https://github.com/spring/uberserver/is ... t-64181239
The problem is most of us have real life jobs, and there's so many things that also need to be done, that we have to prioritize. It was still planned, but not a pressing concern imo.
Also, I've never seen any notable contributions on uberserver from the ZK crowd, so I can't keep but wonder what would've been the current state if they actually put some effort into staying with uberserver, and implemented what they felt was needed.
0 x

BaNa
Posts: 1561
Joined: 09 Sep 2007, 21:05

Re: ZK lobby server split is a disaster?

Post by BaNa » 03 Feb 2015, 14:30

Eh, it makes sense from their POV, why should they stay with Spring community if they have built up a big enough playerbase and non-ZK Spring comminty has dwindled? I haven't been active for a while now, never got the ZK mechanic, but there is no point for ill feelings about them leaving, we cannot blame them for BA having less players or none of the other mods really taking off, can we? They put a few metric fucktons of work into their thing, succeeded and now they off.
0 x

Orfelius
Posts: 103
Joined: 17 Nov 2014, 20:57

Re: ZK lobby server split is a disaster?

Post by Orfelius » 04 Feb 2015, 03:20

BA actually has more players than ZK most of the time :roll:
0 x

User avatar
Licho
Zero-K Developer
Posts: 3803
Joined: 19 May 2006, 19:13

Re: ZK lobby server split is a disaster?

Post by Licho » 04 Feb 2015, 08:17

Gajop, performance and features (like force join) were the ultimate reason.
You didn't see any uberserver commits, because uberserver was deleted and new lobby server created from scratch using different protocol.

In tests it couldn't cope with high player counts (it was dying/randomly breaking long before reaching 1000 active in battle players mark). I thought I would patch it, but after discovering that there is no real thread safety over shared data in it, I decided to rather rewrite it from scratch in C#, and it was done.

Second biggest ill of ZK is engine performance, but that will be solved I hope without the need to do a full engine fork as well.

So please stop acting like it's some malicious act from the ZK. It's driven by ZK needs and I wouldn't spend extra week rewriting entire uberserver+clients if it wasn't needed. I'm a lazy person.
0 x

abma
Spring Developer
Posts: 3548
Joined: 01 Jun 2009, 00:08

Re: ZK lobby server split is a disaster?

Post by abma » 04 Feb 2015, 08:57

Licho wrote:You didn't see any uberserver commits, because uberserver was deleted and new lobby server created from scratch using different protocol.
gajop is refering to contributions to the lobby server. there was none:
https://github.com/spring/uberserver/pu ... s%3Aclosed
Licho wrote:In tests it couldn't cope with high player counts (it was dying/randomly breaking long before reaching 1000 active in battle players mark). I thought I would patch it, but after discovering that there is no real thread safety over shared data in it, I decided to rather rewrite it from scratch in C#, and it was done.
heard about GIL? python runs in a single thread. also you were running a modified uberserver, maybe your changes make it worse: it very likely also runs slower on windows.
Second biggest ill of ZK is engine performance, but that will be solved I hope without the need to do a full engine fork as well.
please do so, you seem to know everything better.

Licho wrote:So please stop acting like it's some malicious act from the ZK. It's driven by ZK needs and I wouldn't spend extra week rewriting entire uberserver+clients if it wasn't needed. I'm a lazy person.
you didn't give any feedback about performance problems to uberserver. also some optimizations are still on the todo, i.e.: https://github.com/spring/uberserver/issues/111

also i'm curious: how do you plan to run so many instances of spring-dedicated? a lot of players means a lot spring-dedicated processes.
0 x

gajop
Moderator
Posts: 3023
Joined: 05 Aug 2009, 20:42

Re: ZK lobby server split is a disaster?

Post by gajop » 04 Feb 2015, 09:16

Licho wrote:Gajop, performance and features (like force join) were the ultimate reason.
You didn't see any uberserver commits, because uberserver was deleted and new lobby server created from scratch using different protocol.

In tests it couldn't cope with high player counts (it was dying/randomly breaking long before reaching 1000 active in battle players mark). I thought I would patch it, but after discovering that there is no real thread safety over shared data in it, I decided to rather rewrite it from scratch in C#, and it was done.

So please stop acting like it's some malicious act from the ZK. It's driven by ZK needs and I wouldn't spend extra week rewriting entire uberserver+clients if it wasn't needed. I'm a lazy person.
If you wanted to work in cooperation with Spring, you could have made uberserver thread-safe and optimized server messages. It might have taken you longer than a server rewrite (which I sincerely doubt), but this is certainly doable and will be done in uberserver's future. I certainly think it would've been worth it, as it actually shows consideration for the work other people have done (backwards compatibility), and centralizes future work (it's not like there are that many devs here).

Writing your own implementation that supports the lobby protocol would be OK with me.
Even proposing a whole new lobby protocol and having people agree on it would've been a far, far better choice.

The point is you didn't feel that was worth it -> this was your decision, not ours: we didn't ignore ZK requests, and I have a problem with some people stating that here or on ZK forums. The fact is, it seems to me you only care about ZK's needs: game quality and user count. I don't think "The Spring Project" is really important for you, more likely just a means to an end, and I feel you would jump ship if a better opportunity presented itself.

The other reason I feel bad about this is because I relied on you for providing Spring services (matchmaking), and I avoided implementing my own proposal because you promised to create something usable for all Spring games (claiming that it would take drastically less time than what I wanted to do). That was months ago (maybe even half a year now), when I had a lot more free time, and seeing as I'm busy now, it'll take quite a while for it to see light on the Spring's server unless someone else does it. Seeing as I think that matchmaking and lobby GUIs are the #1 issue for Spring games, this has considerably slowed down Spring's progress in my view, and also made me lose my trust in you. (This lack of trust is on a personal level with you, not the entire ZK dev team - some of which I trust, and many of whom I don't know or whose opinions I am not aware of)
This will affect how wary I'm going to be of any (direct or indirect) future cooperation I might have with you. To give an example: last I checked ZK's Mission Editor was mostly done by KR, but if you took over, I would be far less excited to do https://github.com/gajop/Scenario-Editor-Core/issues/22 or work on a common mission format with ZK's support, because for all I know, due to a potentially unforeseeable change of ZK's need, you might entirely change your format and all my work being for naught.
0 x

User avatar
Licho
Zero-K Developer
Posts: 3803
Joined: 19 May 2006, 19:13

Re: ZK lobby server split is a disaster?

Post by Licho » 04 Feb 2015, 09:18

heard about GIL? python runs in a single thread. also you were running a modified uberserver, maybe your changes make it worse: it very likely also runs slower on windows.
I didn't know about GIL. In tests it was acting really strange though. Not all clients saw same state if changes were happening rapidly. I assumed its the 25 threads + no lock over global structures, if there is GIL it has to have some other reason.
And on windows, select library function was hitting descriptor count limits so was sometimes dropping connections randomly.
please do so, you seem to know everything better.
Wut? Or are you referring to other ZK devs? I'm not involved in ongoing effort to improve engine performance for ZK, Hokomoko created the no lock lua branch and it's performance is on par with 91.0 that's all I know..
also i'm curious: how do you plan to run so many instances of spring-dedicated? a lot of players means a lot spring-dedicated processes.
Spring dedicated is really cheap to run, there is no indication that that would be a bottleneck. If it turned out to be one, we would simply run them on more machines, because those can be distributed unlike the present form of lobby server.
0 x

Locked

Return to “General Discussion”