Fix mistakes. Use a function.

In D983#38437, @Grugnas wrote:

Perhaps this may be a good starting point to compare units of same class in order to see how different weapon wielding units sharing same price-logic perform in game.

I didn't checked again all templates (unrelated: if we want to look deeper, there is perhaps a lot of polishing work for sounds)

In D983#38415, @Grugnas wrote:

I am not totally against such a patch but I simply wanted to point at the fact that, although the current melee cavalry cost is unusual, the higher than average wood cost requires different decisional paths ( e.g. more workers on wood or even the research of the wood gathering technology from storehouse) which increases the decisions doable by the player.

@Grugnas
You didn't understand and you move the discussion elsewhere.
I have seen many cost distributions in all the past alphas (I follow since 2012), I have also in mind the original designed costs. That's not the point.
Here I am following at least a simple policy: "for all type, cav.cost[type] < inf.cost[type]". If cost differentiation between archers, javelinist, swordmen, slingers, pikemen coud/should be done (and it was in the original design), that's another story.

(Don't worry, I didn't misinterpret.)

In rP20345#24072, @Imarok wrote:

Wasn't there a tool, that checks for missing files?

which seems really bad :p
That's strange as I load all templates. I will check that right now.

In rP20345#24067, @Imarok wrote:

I get this:

ERROR: CCacheLoader failed to find archived or source file for: "simulation/templates/template_gaia_flora_tree_fruit.xml"
ERROR: Failed to load parent 'template_gaia_flora_tree_fruit' of entity template 'gaia/flora_tree_apple'
ERROR: Failed to load entity template 'gaia/flora_tree_apple'
ERROR: CCacheLoader failed to find archived or source file for: "simulation/templates/template_gaia_flora_tree_fruit.xml"
ERROR: Failed to load parent 'template_gaia_flora_tree_fruit' of entity template 'gaia/flora_tree_fig'
ERROR: Failed to load entity template 'gaia/flora_tree_fig'
ERROR: CCacheLoader failed to find archived or source file for: "simulation/templates/template_gaia_flora_tree_fruit.xml"
ERROR: Failed to load parent 'template_gaia_flora_tree_fruit' of entity template 'gaia/flora_tree_olive'
ERROR: Failed to load entity template 'gaia/flora_tree_olive'

when starting a game on Botswanan Haven

(AI should be fine as it uses the resourceSupplyType)

refs r3860, r14727, r7976, r9304, r7259 (I don't put the P after r to not spam those old revs).

All is fine. (checked validation, placed all units and tested in game with gathering in case of).

(don't review that, I will edit that thing)

In D757#38226, @bb wrote:

@fatherbushido mind to click accept for the n-1 state? (when its reviewed ofc)

https://trac.wildfiregames.com/wiki/List%3A_Entities%3A_Nature%3A_Flora
https://trac.wildfiregames.com/wiki/List%3A_Entities%3A_Nature%3A_Flora%3A_Plants
https://trac.wildfiregames.com/wiki/List%3A_Entities%3A_Nature%3A_Flora%3A_Trees

In D590#38184, @wraitii wrote:

I think the cleanest solution would be to send a "prepare for Destroy" message in ComponentManager::DestroyComponentSoon() and have CCmpPosition subscribe to that and move itself out of the world there.

Destroying "now" might lead to inefficiency and weird hiccups (GC-like) and I'd rather avoid that.

In those cases, we destroy the entity and we just want it not intefering with the 'world'. (Like we do in other places).

In D834#38159, @temple wrote:

(Just thinking in general, it might be nice to have the tooltips not reference particular values, so that translators wouldn't have to change the strings every time we changed the values. Units have "detailed tooltips" that show their health, attack, armor, etc., so maybe we can do a similar thing with technologies and their modifications. Some players might want to know that the range was increased by exactly 4m, and others players might not.

(It will be more efficient if bb_ reviews it as he commited the other one.)

ok

(It looks like an evidence in the second decade of the 21st century).
But the word 'formation' has to be well redefined (as it's used to mention different things).

Well, I didn't want commit such patches. (Else we can have tons of such patches).
But why not. (It's reasonnable according to the game I plaid)
In the future perhaps those things should be submit to the guy in charge of the balance or to the balance comitee or forum or whateva (Do those things still exist ?).
Ask @Itms

I consider that as not wanted with our attack system.
So I put that at requested changes (but I don't expect changes).
(Even if in the rts of my dreams, the attack system would be different).
The main one which make that irrelevant is that we don't have point attack.

I set request change as in reviewed but not wanted in the current state.

In D845#37799, @temple wrote:

Added a fruit tree template to reduce duplication. Now the food minimap color is only defined twice, once for bush berries and once for fruit trees.

The same commands on the auras folder doesn't change anything.

(perhaps it's because Math.hypot has a more robust implementation against overflow)

In D969#37883, @elexis wrote:

I wasn't aware that the Math.hypot function already exists.
It does not use our optimized / approximation algorithms, so I did a quick performance test:

let count = 10000;
let numbers = [];
for (let i = 0; i < count; ++i)
	numbers.push([randIntInclusive(1, 100), randIntInclusive(1, 100)]);

let start = Engine.GetMicroseconds();
for (let i = 0; i < count; ++i)
	Math.hypot(...numbers[i]);
warn(Engine.GetMicroseconds() - start);

start = Engine.GetMicroseconds();
for (let i = 0; i < count; ++i)
	Math.euclidDistance2D(...numbers[i]);
warn(Engine.GetMicroseconds() - start);

WARNING: 9694
WARNING: 8378

So it seems to be slightly slower.

(random, perhaps useless or irrelevant thought when reading)

(one can also nuke it, or make it all mod compatible)
(enabling it by default and only displaying error is also an option but it could be bad for other mods)

empty line

closed by rP20314

@wraitii :
I just noticed that the check for BuildRestriction seems not being strong enough (exact opposite of training restriction).

In D945#37549, @Mate-86 wrote:

Is there someone needed from C++ experts to give the green light for such ideas or should I just freely work on the code and we will see? :)

fixed in rP20313

All that is beautiful!

Your idea sounds ok.
(Basically move that part in the EntityLimits component)

In D945#37521, @wraitii wrote:

How are flames created by flaming projectiles handled right now?

nice teaser!

(fixed in rP20303)

(just for ref: https://code.wildfiregames.com/D921#36217)

(We start digressing but there is a patch lying somewhere since years with the ground base for nuking the short pathfinder and for simple steering behaviors).

Indeed, it's bad to initiate those queries and to not use them.
Tested with female inspiration tech. Nothing wrong.
Looking at the code, I don't expect any wrong thing. (I tried to imagine some weird cases).

All that is beautiful.

Patch works as expected (and is indeed needed)
I didn't look at the code (at least I didn't see something you could have miss). If you need a more detailed review, I can look at it.
You need to update (maintain) the Auras test
(adding Engine.LoadComponentScript("interfaces/RangeVisualization.js"); should be enough)

fixes

I didn't remember. I guess it's better like that.

Asking that could have the contrary effect as the one expected...
EDIT: To be more explicit, it's not the place to ask such things. Sometimes it's better to not have answers. Sometimes it's better to let things rip.

You're a fast reader.

It's usually hard for me to tell what you mean, if you're making fun of me or something else. :)

After reading threads and tickets and patches over the last week,

You're a fast reader.

I think I'd rather work towards adding something like steering behavior, possibly even replacing the short pathfinder.

That's what was/is planned, and was always in mind (already in 2010). (explictily or not)

(Of course I'm sure steering will have its own issues.)

As you know Steering is broad.

But I'd appreciate advice about this.

Those things are 'easy' outside 0ad. In that specific context, you will need a global view on what do we mean by group walking, by formation and things like that.

And I think it would make formation movement nicer too.

Depending of what we mean by formation, I'd say replace?

In rP20263#23659, @elexis wrote:

Mostly didn't understand your comment then. How did you test? Which function did you use to measure the time? Date.now()?

In rP20263#23656, @elexis wrote:

The link you sent isn't relevent because we have a custom Math.pow function. Also browsers can have a custom implementation. Testing it in a browser is more relevant to webdev.

You didn't read what I wrote.

(I don't know how that works internally but the change about removing cos = Math.cos(a) doesn't seem changing perf.)

(if you want you can rename vector.js in Vector.js (we decided that somewhere, but I was always afraid to do that)).

I permit myself to add a com (as phab sent me a O2 thing that morning).
I post that here instead of D950
Nothing critical but:

• I guess the current code was done like that (mult instead of Math.pow) for performance issue.
• We could just ignore that as we don't really take care about that kind of things in some other places (me the first).
• JsPerf report really various diff (it seems that most recent interpreter crush those diffs) https://jsperf.com/math-pow-vs-simple-multiplication
• In my own test (with many biases as time precision and things like that), I noticed that the new one distanceToSquared (in our engine, not in a browser) has 3x running time on average.

In D757#37037, @bb wrote:

Not exactly just "Javelin" or "Archer" would be enough (but putting "Ranged" before is allowed, but it is only a name). What I have in patch is merging the current 4 attack types into 1, which have all the features we have now and even some more, f.e. we can have projectiles with capture "damage" (as in @Stan suggestion for throwing rocks to break windows in capture). Probably its a bit vague what I mean here, but I will come up with a patch in a bit (mostly need to rebase).

Ok for n-1 (with attack_impact_type).
Unrelated to this patch, for your bow and javelin thing, you mean you would have Ranged_Bow and Ranged_Javelin and something like a start with "Ranged" in the PerformAttack code?

(one could perhaps also do things in MT_PositionChanged)

@bb : I didn't request that at all, my pov was that attack_impact sound are a bit meaningless for other attacks so you don't add to add that code :) For example for capture, you would hear the capture attack sound (currently there are none I think) then you will immediatly hear on each tick a capture attack impact sound, that 'sounds' really weird. I have a similar doubt for the melee attack. That was just something I was thinking about (should we change attack_impact to attack_impact_type?) But your patch, your choice ;-)
So: I will review/'accept' version (n-1) or that one or version (n-1) withtout attack_impact_type (my own pref is 3).

In D757#37005, @Stan wrote:

In this case maybe it would work for hybrid attacks, like AOE attacks.

In D757#37001, @Stan wrote:

Mêlée makes sense for slashing sounds ranged too. Capture could make sense if it's some kind of cheering and breaking glass sounds.

I wonder if attack_impact make senses for all the kind of attack?