Feedbacks from A26 SVN tests

[wowgetoffyourcellphone]

7 hours ago, faction02 said:

The Han have also access to will-to-fight on top of their own fort technology when they capture an enemy fort.

 

Fixing this now. Thanks for the report.

 

7 hours ago, faction02 said:

I tested the Han, I noticed that they have access champions through captured stables or barracks, not sure if that would make sense since they don't have to search for some upgrades or anything of the kind.

 

I have ideas on how to fix this, but it will be slightly messy on the backend and a little hacky.

 

7 hours ago, faction02 said:

After capturing an enemy fort and its stables and making all upgrades and getting the right hero, I was able to spam champion crossbowmen war chariots that had 114 damages per 1 second. That seemed to be slightly too high compared to the already quite good performance of britanic war chariots with Boudica that can currently reached 85.7 per 1.25 seconds ...  

Can you step by step this for me? It would help to reproduce and create a fix.  

[faction02]

After seeing the super-champions, I was disappointed not to find any starships available for the Han.

 

The champions start at 40/3sec

First attack upgrade => 46/3 sec

Second attack upgrade =>52.9/3sec

Third attack upgrade =>63.5/3sec

Poisonned arrows => 76.2/3sec

Repeating cross-bow => 76.2/2sec

Hero Wei Quin => 91.4/2sec

Cross-bow training => 91.4/1sec

Will-to-fight => 114.6/1sec 

At that point champions palace guard have only 31.2/1sec  whereas citizen-soldiers crossbowmen gets to 47.6/1sec.

Both crossbow training and repeating crossbow reduce by 1second the firing time of crossbowmen. That’s 2 very powerfull upgrades.

 

Wei Quin is also a bit too powerfull with respect to Boudica. It seems there is still to decide which bonus to give him. He gives some bonus to champions as Boudica, + a malus to enemy cavalry + he has a flame thrower. Making 10 damages independently of the level of armor can be significant in some cases. Pikemen are the only melee unit currently immune to his flame thrower because of their extra range.

 

Not sure it is meant as a feature, but after running the upgrade « imperial court », it is possible to train champions guard (no cavalry champions), but no longer possible to train citizen-soldiers from the cc. The building to train all champions costs 300 stones/300 metal, but the upgrade to transform the cc into champion building costs 300wood/300stones. Considering that other civilization pay 600 food to get champions at barracks or stable and remain able to train citizen soldiers there, that sounds a bit too much (seleucids do not even pay anything for that). I would probably go for the extra building in all cases and never research the upgrade since during the 40 seconds of research time, the cc can't be used to produce units.

 

The Temple and the LaoziGate health regeneration capacity seem to cumulate (both healing the same units at the same time). Not sure that should be desirable. I couldn't test if the LaoziGate would work on ally, but setting the range to 50meters for the LaoziGate would at least be consistent with the revered monument of the iberian range.

 

[wowgetoffyourcellphone]

Excellent feedback on the Han. I'll have some ideas tonight on how to address them.

I may do some reading and revamp the heroes a little. 

The intention was indeed to have the Imperial Court not train citizens anymore. Perhaps the function of the upgrade can be revisited. Perhaps the Imperial Court unlocks heroes there but also increases the speed of the entire production queue substantially so that training and research there is super fast. Then we can limit the upgrade to just 1 civic center. 

[Nullus]

10 hours ago, vladislavbelov said:

I added a fix, though it might have a bit different stack. Could you try the latest SVN?

I don't get the error anymore, thank you!

[real_tabasco_sauce]

5 hours ago, faction02 said:

114.6/1sec

this is also with greater range than briton champ chariots and firecav. So Omega OP. 

Three unique upgrades that apply to crossbows both CS and champs seems like an impossible balancing situation to be honest.

Either: 1. you balance for the upgraded state and the units are only good after you get all those upgrades or 2. you balance for the unupgraded state as above and have godlike champs for the late-late game that nobody can beat.

balancing between 1 and 2 would be ok i guess, but i think these unique upgrades would work better as tradeoffs of sorts. (increasing cost, or movement speed, or accuracy as examples)

Edited March 2, 2022 by real_tabasco_sauce

[Freagarach]

22 hours ago, andy5995 said:

Bug: Near the end of the game, Han Chinese enemy building gets destroyed infinite number of times and game doesn't end.

Video clip on Twitch

 

Using self-build at git (GitHub) commit e723cc4

 

mods enabled: community maps 2, v0.25.11

 

commands.txt 325 kB · 2 downloads metadata.json 74 kB · 2 downloads

Isn't this just the AI keeping trying to construct a structure at the same location?

[wowgetoffyourcellphone]

7 minutes ago, Freagarach said:

Isn't this just the AI keeping trying to construct a structure at the same location?

Yes, still a bug, just not a "Han" bug.

[wowgetoffyourcellphone]

7 hours ago, faction02 said:

Will-to-fight => 114.6/1sec 

 

Fixed here: https://trac.wildfiregames.com/changeset/26516#file1

 

1 hour ago, real_tabasco_sauce said:

balancing between 1 and 2 would be ok i guess, but i think these unique upgrades would work better as tradeoffs of sorts. (increasing cost, or movement speed, or accuracy as examples)

"Repeating Crossbows" looks to be historically problematic, only becoming a military application in the slightly later Tang dynasty. I'm removing it, so that helps the situation.

Also, I'm changing "Crossbow Training":

Old: "Crossbow Infantry and Cavalry −1 second firing time −20% train time."

• This is essentially a -33.33% firing time reduction.

 

New: "Crossbow Infantry and Cavalry −20% firing time −20% train time."

 

[wowgetoffyourcellphone]

@Han Crossbow teching: https://trac.wildfiregames.com/changeset/26519

[faction02]

7 hours ago, wowgetoffyourcellphone said:

The intention was indeed to have the Imperial Court not train citizens anymore. Perhaps the function of the upgrade can be revisited. Perhaps the Imperial Court unlocks heroes there but also increases the speed of the entire production queue substantially so that training and research there is super fast. Then we can limit the upgrade to just 1 civic center. 

I do like the idea of that upgrade a lot in its principle, it forces to make a choice. Maybe that upgrade might still be worth it if someone wants to mass champion guards, in which case he would get the upgrade on top of the other buildings for champions production.

Comparing to the Mauryans, their palace which trains their guards costs only 200 stones and 200 metal. Everyone get at least one palace anyway to get access to heroes. However, I don't have the impression that these champions are that much used despite the opportunity cost of producing them being essentially zero. If the current balance do not change much, there is probably no emergency and no need to think too much over this comment for the next alpha.

Something like having a hero garrisoned in the cc to substitute for the upgrade could be interesting too if there is some historical justifications for the Imperial Court to be directly linked to one of the heroes. The hero garrisoned in the cc would then unlock a fast champions production for that building. The opportunity cost of not using a hero on the battle field can be important so one would have to choose whether he wants to keep raising more champions or use the military advantage provided by the hero. 

[Philip the Swaggerless]

On 27/02/2022 at 4:08 AM, LetswaveaBook said:

I think the acceleration also looks wierd. It looks like there is an invisible man the cavalry pushing (or pulling) them forward until they reach maximum speed. I think there would be a very improved visual experience if the animations between the acceleration and full speed phases where different.

Maybe scaling the length of the run animation with the units current speed would make it look very decent.

Hello.  I downloaded and played SVN single player yesterday.  I think the acceleration looks like a bumper car.  I think it would be better if the units didn't begin accelerating from zero, but move at half their normal top movement speed (not chase speed) to begin with and accelerate from their.  Or perhaps ALL units could start at the same speed, 5, if the intention is to make it harder for cavalry to disengage.  That's probably better.  And most acceleration should happen at the beginning.

[Old Roman]

 Just updated SVN to Revision 26523  Error stream starts on startup

Can load match but no control over movement of population

Only minimap visible

Windows 10

 

[bb_]

Wait for the autobuild tomorrow morning, or revert to rP26521.

[andy5995]

The build appears broken on Manjaro 21.2.4

Quote

Environment.cpp
LightControl.cpp
In file included from /usr/include/c++/11.2.0/cmath:45,
                 from /usr/include/c++/11.2.0/math.h:36,
                 from /usr/include/wx-3.0/wx/math.h:29,
                 from /usr/include/wx-3.0/wx/gdicmn.h:23,
                 from /usr/include/wx-3.0/wx/event.h:20,
                 from /usr/include/wx-3.0/wx/wx.h:24,
                 from ../../../source/tools/atlas/AtlasUI/Misc/precompiled.h:107:
../../../source/tools/atlas/AtlasUI/ScenarioEditor/Sections/Environment/Environment.cpp:32:13: error: expected unqualified-id before numeric constant
   32 | const float M_PIf = 3.14159265f;
      |             ^~~~~
make[1]: *** [AtlasUI.make:315: obj/AtlasUI_Release/Environment.o] Error 1
make[1]: *** Waiting for unfinished jobs....
make: *** [Makefile:181: AtlasUI] Error 2
make: Leaving directory '/home/andy/src/0ad/build/workspaces/gcc'

Latest update to Manjaro updated gcc to 11.2.0 and glibc to 2.35

 

At git commit d25309b (Wed Mar 2 05:00:23 2022 +0100)

[vladislavbelov]

47 minutes ago, andy5995 said:

Latest update to Manjaro updated gcc to 11.2.0 and glibc to 2.35

glibc added a new macro M_PIf to workaround the bug few months ago, I'll fix that tomorrow.

[vladislavbelov]

20 hours ago, andy5995 said:

Latest update to Manjaro updated gcc to 11.2.0 and glibc to 2.35

Fixed in rP26536, thanks for the report!

[ValihrAnt]

1) When hovering outside of home territory with a building selected, which is not meant to be built in neutral, receive errors.

ERROR: Parameter without value at pos 74 'House cannot be built in [object Object] territory. Valid territories: own'

ERROR: Invalid tag 'object' at 75 in 'House cannot be built in [object Object] territory. Valid territories: own'

2) When trying to autocomplete in a match lobby get errors. In game and in main lobby works fine.

3) When hosting without STUN get warnings and others can't join. I consistently host games in a25 lobby without STUN without problems.

2) and 3) shown in the interestinglog.html

interestinglog.html

[Stan`]

Thanks @ValihrAnt 
1) Was fixed today.

2) @bb_might know. cc @Silier

3) Is weird we didn't change anything. Maybe a bug on the lobby @Dunedan , @user1

[bb_]

2) https://code.wildfiregames.com/rP26438

[faction02]

Units movements feels currently very "weird" . It is quite difficult to describe exactly the issues with units movements since they seem to take many forms. I am having concerns that the introduction of acceleration, while units pathfinding is still a task under development, might just be too early.

I would guess that playing with some numbers might help to reduce the weirdness of units movements, but I was wondering whether it is not also making pathfinding problems more difficult to solve with potentially some impact on the game performance?

[Stan`]

45 minutes ago, faction02 said:

Units movements feels currently very "weird" . It is quite difficult to describe exactly the issues with units movements since they seem to take many forms. I am having concerns that the introduction of acceleration, while units pathfinding is still a task under development, might just be too early.

I would guess that playing with some numbers might help to reduce the weirdness of units movements, but I was wondering whether it is not also making pathfinding problems more difficult to solve with potentially some impact on the game performance?

There is a patch for it. https://code.wildfiregames.com/D4512 You might want to voice you opinion there. @bb_ might have more insights on the pathfinder and performance. This is for me a controversial patch (realism vs gameplay) so I do not have a strong opinion on it. I'll listen to the numbers I suppose.

[Old Roman]

This might have been brought up awhile ago, but the save function in the current build (26579)  and earlier builds completely freezes the game.

[Stan`]

7 minutes ago, Old Roman said:

This might have been brought up awhile ago, but the save function in the current build (26579)  and earlier builds completely freezes the game.

What if you delete your saves?

[Old Roman]

Yeah, I remember reading that being the solution, but I was looking for a  saves folder within the SVN folder.

Deleting all the saves generated in 25b in the main folder solved the problem!

Thanks.

[Salesome]

The 0 A.D. snap development package crashes when I try to start an actual game match. The snap is currenty on revision number 26557.

After launching the game, in the menu, an error message pops up (see attached screenshot).

When I try to start a single player match, the game crashes (see crashlog, also attached).

I found the forum thread below where an identical error message is discussed. I double-checked and there are only standard ASCII characters in the paths for the installation on my laptop.

The current alpha 25b snap does not show these problems. OS is Ubuntu 21.10.

crashlog.txt