Spyhawk

@Mister J Before this happened, were you swapped from a team to another one by any chance? We fixed an issue related to weapon recently in master.

For the crash, we'll need the log. Whoever's in charge (@MiniMuppet?), if you please...

The !listplayers issue is known, but this is from WoldAdmin side, which isn't maintained at the moment. You can use the built-in `/players` command as a workaround.

21 hours ago, Svepa said:

I believe I’ve finally figured out an issue I have when I have panzer.

I’ll be using the panzer, die and respawn as a soldier with the default machine gun and that’s it, I’ll the notice someone in my team is using a panzer, so it looks like they’ve stolen it.

I think it’s a bug, but basically I believe what happens is I die, then another soldier on my team sees a panzer on the floor and picks it up, then when I next spawn and notice I’ve been stripped of owning the panzer I have to re-select my weapon of choice in the menu.

I hope this makes sense, but would be great to fix this

I'd say this is more a limitation of the weapon restriction feature than a bug, and that it shows more easily on crowded server. I wouldn't know how to fix that properly: preventing players to pick a weapon laying on the ground because its "owner" is in limbo doesn't sound great to me.

Thanks, looking into this. Keep the feedback coming!

Changelog for this new version:

Quote

• Fixed simpleItem colour not updated for current weapon when picking up weapon

• Allow spectators to send team messages to each other

• Fixed max chat length in intermission

• Fixed dynamic maximum health computation

• Added cg_etVersion info display in /players

• Fixed stats of previous map are displayed in warmup

And Mac OS X files are back!

@MiniMuppetThe previous build doesn't have the mac files, due to a last minute issue with that platform. I just sent you a PM regarding a new build, which does include mac files. So mac OS X users will be able to play as soon as you update to the new version.

(we asked Eagle to update to a new version with mac files a few days ago, but I guess he couldn't do it).

Yes, this is working as intended. Some mods do take gib into account (like NQ - which counts spent bullets), but we especially don't want to "punish" players when they gib by computing a lower accuracy. Rather, we prefer to display the real accuracy on alive target, like many other mods do (such as etpro).

Issue 1/ is now fixed in master, and 3/ is a misconfiguration by TM. 

The team_maxAirstrikes and team_maxArtillery cvars control the arrow below the chargebar, but if I understand @Duro correctly it is the chargebar itself that seems slower. The chargebar is player dependent. You need both (player chargebar full, and team arrow "ready") to have a fieldop call a strike.

Might be a TM or even map custom setting, but from what I can see, if it times at 40 sec (without CP built), then it's behaving exactly as originally designed. I don't see anything strange on my local test.

4 hours ago, Duro said:

3. Since the update Fops bar is recovering very slowly

What do you mean by "very slowly"? The default value of g_fieldopsChargeTime is 40 sec, but the actual time can be shorter if a CP has been built (25-30 sec).

Issue 1/ is related to a recent bugfix. We're looking into 2/, thx for the report.

20 minutes ago, Charlie said:

A minor detail but !listplayers wasn't working fully in between these crashes listed above. It would only list a dozen or so players until the server "crashed" again and it corrected. I didn't get to test other commands out. I would hazard a guess that something is interfering or conflicting with wolfadmin or something related to it? 

I think that issue has been reported before. Timo is the one handling WolfAdmin dev (it's not directly related to ETL), but he's kinda very busy currently.

You might want to check who's client 16. We fixed the UTF8 issue, but maybe we missed something here. Anything special in nickname?

We're also adding extra debug information to track the MSG_Copy crash, that will be in the next build.

@SvepaIt's fixed in master now, and will be available in the next build. Note however that already sent message (before intermission) won't be adjusted, only those sent during intermission. 

Some long time regression should be fixed: the UTF8 issue with vanilla clients connected, the weapon disappearing after revive, SR being suddenly reset to zero (not 100% sure about that one, but we added extra debug information to help fix it in case it does happen again). If any of the mentioned issue happens again, please inform us, otherwise we'll consider them - hopefully - definitely fixed.

33 minutes ago, Spyhawk said:

And hotfix coming shortly.

And hotfix coming shortly.

There are some SR fixes in the next snapshot, but that one isn't on TM yet.

15 hours ago, Ctrz said:

I understand that. But I can only see SR ﹥ 0.00 . Would be more helpful to see if player is for example -0.04SR overall or -27.04 .

Ok, I see your concern now. As a general rule, player skill is modeled as a Gaussian, and SR values below 0 are extremely near that value. There isn't -27.04, more like -0.02. Think of it as a value that is slightly aggressively rounded.

As a sidenote, the previous old versions (the ones that didn't crash) had some issue with the old vanilla client that might cause some value to be surprisingly set to 0, so some fixes have been added to avoid some weird edge cases. That issue need monitoring to check if the fixes completely solve the issue in the new version, or if more additional work is needed.

18 hours ago, Ctrz said:

Yea. Ive seen it. In past, I have joined game - map ends in 3 sec, I don nothing and gain 0.03. Then other map I also play and my team win and i get - SR from that.

Just that those 0.00 players dont say anything to me when we try to see if teams are even or not (if we regard SR as the measuring tool) because one can be -0.01 and other one -6.22. There lies big difference if you compare them. Bigger SR usually says for me, that player knows the game, the map and has some good skills (in shooting, in engineerin, in obj stealing/defending - whatever) because you cannot have positive and good (lets say above 10?) SR when you are simply carried around by a strong team.

Its a good analytical tool and makes some sense but not when we dont see full values of all participants.

I'm honestly not sure about what you're trying to say here. If you're saying that people try to gain SR by simply jumping in the winning team in the last seconds, yes, their SR data will eventually increase (through the opposite is possible) but that is statistically irrelevant (statistically is a key word here). That won't help anyone, unless they do it for literally hundred or thousand times. The UI is actually limited so only 2 digits are shown, but under the hood it is much more precise (5 digit if I recall). Anyway, take SR as a very conservative estimate of relative player ability to help winning the map that is inaccurate at a specific point in time, but that eventually converges towards a good estimate on the longer term.

Also, you do see SR values of each players in game, in the scoreboard. In-game ranks are tied to SR values, so you can get sense of who has high SR by looking at their rank icons.

Any value could be used, but we use the standard TrueSkill values in Legacy: a Gaussian going from more or less 0 to 50, centered on 25. Don't focus too much on these negative values, the important bit is that negative values are smaller than 0 or 25. Shuffle handles it correctly, since it does "smaller than/greater than" comparison atm.

To answer your question: you can't know how many maps you need to play, since your score depends not only on your performance, but also on your team and the opposite team performance. The fastest way to improve your rating is to play in the weaker team and win the map, as the algorithm will favour the underdogs. Winning in a very strong team will actually not help much.

SR is an implementation of TrueSkill, a Bayesian inference algorithm. It has nothing to do with kill and death, but takes win and loss into account, as well as team performance and time played in each team. It has also been adjusted to add map bias as a parameter. This is why people that switch team, go to spec or disconnect/reconnect before the end of the map can't escape it.

We're investigating that crash, but logs didn't bring any info. We're actually unsure if the new version is actually related (keep the log coming).

Also, please tell us if the "weapon disappears after being revived" bug happens again. It was hard to track down, but should be entirely fixed in this version, so feedback appreciated here.

What's the timer value when the issue occurs? any idea?

Never seen that issue. Is that a consistent issue? Does somebody else crash near the end of round?

`0` has the same effect as `1`. So `class m 0 0` is the very same as `class m 1 1`.

As a general rule, use 1 for pistol only, 2 for akimbo.

And FYI, Legacy has also an in-built class selector too if you don't mind it being a little bit slower to use than custom binds ("P" key by default). It doesn't allow to select secondary weapon yet (pistol/akimbo), but we'll hopefully extend that soon. Team can be selected with the "J" key.