I think current rs is ok. It looks fine. Not too fast and not too slow.

Fighting in formation with defensive stance looks like in svn (broken, but not more as it is now).
Formation does not reshape while doing stuff.

reflag

oh, my bad :)

From the sixth century BC ...
It was from their Iberian neighbours that in the fourth century they learned the technology for turning pottery on the wheel and for making rotary querns to mill their grain. A century later a Celtiberian script had been developed

The Ancient Celts (Barry Cunliffe) page 165

Gameplay:
It makes sense that when one captures some mercenary building, one is able to train that mercenaries as they are not of the civilization that constructed the building.
For nuba village and blemmye camp see comments from ValihrAnt and brog-.
Other:
Trained units keep ability to build structures iff player has that structures in the roster and was able to do so without this patch.

@wraitii needs to accept it first, then @borg- can close it

Well I did not my homework too good,

Thanks for the clarification. What determines then the space where clicking would select an entitiy? The footprint?

Visual component so entity mesh.

Is <Clearance> in simulation/data/pathfinder.xml the unit obstruction size then?

Yes, it is.

By saying how selection is done I meant it does not define if mouse pointer is actually hitting unit so it is selected. It just defines if it can be selected at all.

use the <VisualActor/Actor> for calculating the hit area?

Simulation should be independent of graphic and art so no.

keep <Footprint> and <Selectable> as is, but introduce a separate <HitBox> node

That would be more problematic because it would mean melee units will hit places which ranged not as melee move based on obstructions.

However, currently the footprint is the area an entity occupies and the <Selectable> node is its visualization.

No it is not by default. Binding between selectable and footprint was introduced in rP7479 to better mimic shape of entity. It was just constant circle until then.
Also area that entity occupies is not very correct. Entity has obstruction and that area is occupied by entity and cannot be build on and is collided when moving.
Footprint is just hitbox for ranged attacks and unit spawning to spawn nicely since their obstruction is too small and they would one over another when spawned.

Furthermore, what happens if two entities have footprints that don't overlap, but the <Selectable> area does, and you click on the overlapping part? Which entity is selected?

This is not changing how selection works, just how it is visualised. Selectable component has nothing to do with fact how selection is done.

Also, shouldn't the circle or square be moved inside or merged with the <Overlay>?

Hmm, yes it probably should.

And does your implementation work with entities that use an <Outline> instead of a <Texture>

I did not change how are things rendered. Just allowed to read data which overwrite data from footprint. Combinations among Square/Circle and Otline/Texture works just as if defined only in footprint.
And yes we can even question now if it would not be better to define that shapes in selection for selection purposes, but just to allow to have crazy footprints for selection without affecting gameplay this is enough to do without need to edit every second template. ;).

I think 1 or 2 can be enough, just to not have it for free.

create square formation of 35 units. Set them to defensive stance. After arriving to position some will move to the Centre of formation.

I think this feature can in some cases mess up ai planning but it is similar if entity would upgrade or get deleted. as far as it will not be used to remove crutial things it ought to be ok (like removing or adding Siege class to units in the battle not sure how that would play out)

filter
fix tests
fix naming
remove redundant macros

It is nice idea, but one should do profiling at some point.
Adding another range query is not for free :).
Although I do not expect big perf issue because of that.

I do not think it is so slower. You can do stabbing with spear as fast as with short sword when staying at one spot.
And swordsman does have 750 repeat time so this is still slower.

Also might not be important, but in theory if game is won, winning players should be able to split remaining pop cap. But that is nit-picking.

If one plays with hacked popcap (gift from the gods e.g.) and any player is defeated/resigned cheat influence on pop cap is removed.

I can agree with allowing to build them on neutral territory and removing territory influence.

Also removing special kush buildings from building list, I do not agree with that one.
Replace kush -> {civ} to avoid construction by forbidden civilization and to keep for kush.

I believe one can use {native} tag to apply civ code from structure and not owning player so one can avoid renaming units.

fix tests one or another way

rebase
mark mod as invalid and display in list as disabled and display reason instead description to not spam ugly error messages on screen as this is not error by the game but of the moder
report failed signatures back to list of mods
fail if property is not set using strict mode when getting from js
check in js for undefined values

I would not say. Being in cheering state and not playing cheering animation is bad as it appears to be idle and it is not.

I actaully was thinking to make cheering as template option, if unit should cheer at all. E.g. ram, ship, animals ?
But then it should probably tell friendlies to cheer or not.

working version
fixed wrapping and clipping

@vladislavbelov would you have idea why it splits placeholder text to multiline by empty spaces ?

rebase
remove inheritance

found issue, that it will split longer text to multiple lines based on spaces

Attack damage is lowered as nerf.

@Nescio do you want adjust donkey trader footprint or it is final ?

They do not die doing nothing. Actually they do what in svn so ok, not braking that part.
They do not reshape when hunting or fighting (what was the goal of the diff).
After hunt in defensive stance with twin formations it is a bit mess, but not related to diff directly as it is related to merging twin formations, also sometimes they end up not merging, what currently is issue as in replay they ended fairly close to merge ( at least I guess they are in merge distance ).
(not merging replay)

you have broken unit_ai tests

if windows, you need to rebuild, there have not been autobuild yet

Also this will happen with defensive stance told to hunt (no formation) target in svn.

This doesn't seem to be systematic. See

I suspect target is too close to cc to see the effect. But if it is behaviour of single unit to stop at resource dropsite, ok I do not argue it is wrong to stop there I literally gave question with the pictures. Anyway one or another defensive (svn or diff) behaviour of formation is not correct as they clearly do not behave the same hence I considered new one as bug.

Held position is not updated in defensive stance so targets around are too much far away, they do not react to attacks.

Without diff they stop roughly around dead target after gathering.

So you are saying this behaviour is correct not reforming at the position they would stop without formation that would be position of killed target.

Twin formation hunting. It looks very broken with defensive stance.

Issue reported in ticket cannot be reproduced with this diff.
Removal of FindInPosition is not problematic.

It is at the line I already said.

I broke it.
As I said earlier.

Indeed, linear damage is there to demonstrate power of the shot as it would damage more targets behind current one.

static casts and years pls :)

hmm, that seems like it's trying to close again when already closed, which shouldn't happen. Perhaps it's an art issue as we end up resetting some animation timer. I'll check.

did not test it but looks good by checking the code

garrisoned aura

Are the donkey's hindlegs inside?