Think you forgot tests :)

Build is green

Logic error.

Build is green

Check for cmpGarrisonable before using it.

I could see this being very useful in differentiating some civs. Perhaps Romans, for example, only trains soldiers in these groups. Or maybe Champions train in groups. A group can include an officer (2x health, attack aura) and a bannerman (3x health, speed and health aura).

Maybe although if there are a lot more files might make more sense

Seems like a more useful place. But maybe you should rename the files to simd.cpp/h for more general use and preparation for future usage of AVX/AVX2?

Build has FAILED

Build is green

Build is green

Build is green

Build is green

Oh yeah this is almost a duplicate of that, in fact. I've posted some of my comments there, I'll change this bit here to be about upgrading wall turrets into proper towers then after that diff goes in.

I failed to realised that this existed, but I fully agree with this and duplicate it somewhat in D3466.

Build is green

So step one would be D2769.

This looks OK to me, but haven't tested, make sure to check it still works after the changes (and forgot about const, sorry)

more comments

Build is green

fix comments

Hm, indeed, I was wrong in that we need the object anyways for melee for attack effects.

So you would say rather two extra properties for D781?

actually reuse already translated string

Build is green

reuse string

Build is green

Build is green

restore counter after correct password so user can misstype when rejoining

Build is green

Build is green

Build is green

I see no reason to reduce health to 1000. In general, they are bigger than a town tower and also the walls. Decreasing the max range / vision is bad because they are the same height as a tower. What I would do is add garrison attack/vision depends of units like a outpost, decrease health by 10-15%, increase stone cost and construction time.

After discussing with FeldFeld, only do the HP change for A24.

Fixes: #5948

Build was aborted

I don't think the added object is a huge problem, but I agree that this currently seems like change for change's sake.

As angen said it adds a few object creations... It would remove the addition of parameters in the future though. Maybe it should be profiled.

without reading just that creates another object

I had a thought on this, basically 'rounding for humans':

• As an upper limit, we never really need more than 2 significant digits for the decimal places: 0.25 is better than 0.246
• However, we probably want 2 significant digits at least: 0.25 is also better than 0.3 or 0.2 (see gather rates).
• But we're not scientists and we don't care about 'insignificant' significant digits: 0.1 is generally better than 0.10, a fortiori 1 is better than 1.00
  • This is debatable to an extent, and we might prefer 1.0 over 1 or 1.00 if all other related numbers have decimal places, just for visual coherence.
• We probably don't want to round integers: 12455 over 12000, even though it's also 2 significant digits.
• But we probably no longer care about decimals at that point: 124 over 124.2
  • I would say above 10, only 1 decimal place at most instead of 2, and above 100, none: 1.23 over 1.2, 12.6 over 12.57.
• Multiples of .25 are probably readable enough compared to rounding to the upper/lower digit, e.g. 3.75 over 3.8 even in a "1 digit" context. Likewise, 24.25 over 24.3
  • Also debatable

Any conceptual objections against this patch, @Angen, @bb, @wraitii?

Rebased.

Rebased (in anticipation of D3285).

Build is green

Rebased in anticipation of D3463.

(Refs. D3289.)

yes we should but I was lazy to start that, see attached comments in https://code.wildfiregames.com/D2053

In D3464#152961, @Freagarach wrote:

Why 10 instead of 8?

Why 10 instead of 8?

I think we should probably change that as part of a more extensive stance rewrite ;)

I would actually like to just remove those lines. I frequently order my units somewhere and thereafter set them to standground. But I guess I might be alone on that ;)
For now this makes sense more (A23b behaviour).

We should have a function for max decimals I guess.

This writes 1 as 1.00 but I think it's actually OK for readability because all numbers use the same decimals.

Two is fine for me, I was merely asking ;)

In D3460#152895, @Freagarach wrote:

Why 2 and not 1?

Why 2 and not 1?

Build is green

Pics ?