JS only.

Successful build - Chance fights ever on the side of the prudent.

Successful build - Chance fights ever on the side of the prudent.

C++ changes only, CC

In D4640#197298, @wraitii wrote:

you're also going to render them in that case, despite the width/height being 0 (unless I'm incorrect here) ?

In D4640#197295, @vladislavbelov wrote:

In D4640#197284, @wraitii wrote:

You might run in a situation where clipart == CRect(), but I'm not sure that's actually something that can happen in practice.

Currently it's intended to not clip if clipart == CRect().

In D4640#197284, @wraitii wrote:

You might run in a situation where clipart == CRect(), but I'm not sure that's actually something that can happen in practice.

For reference, AoE 2 added a ranged/melee unit recently: https://www.youtube.com/watch?v=rvJdOKktoMA
The switching is not automatic but manual.

In D4640#197283, @vladislavbelov wrote:

In D4640#197280, @wraitii wrote:

not clipping if the clipping area is 0 by 0

Why?

In D4640#197280, @wraitii wrote:

not clipping if the clipping area is 0 by 0

Why?

As I pointed out in the forum, 0 and undefined aren't actually equivalent anyways. The formula we use is:
effectData.Damage[type] * Math.pow(0.9, resistanceStrengths.Damage[type]). You'll notice that Math.pow(0.9, 0) is 1, whereas Math.pow(0.9, undefined) is NaN, as one would expect. Thus a guard must be added, and thus there is an obvious difference.

With the inline and D4605, I get no OOS with the rejointest.
The patch reads correct from a functional view.

The pass class name _was_ serialised? And the reason it worked was probably because the pass class name didn't change ove the course of a match. ^^

In D4486#197134, @wraitii wrote:

The specific approach taken here doesn't strike me as bad, but I think C++ wise I'd prefer what I've done in D4630 with the component manager. It's less code & more maintainable IMO.

If you could fix/merge the convenience function please, then I can move that from this diff and go back to JS. ^^

see D4639

Arcanist wouldn't let me apply the patch, but doing it manually did the trick.
Patch fixes the problem described in the ticket.

Successful build - Chance fights ever on the side of the prudent.

Successful build - Chance fights ever on the side of the prudent.

a problem with openGL ARB was tracked down to this changeset.
see #6537

I suppose comparison with older versions is a bit unfair due to the very badly optimized greek helmets...

Apply the patch and compile the game

Ok

Compare performance for Combat Demo Huge

Before: ~170
After:~85
https://ttm.sh/bkP.jpg

Could be nice to have tests for this class if possible? Using the test mods.

Successful build - Chance fights ever on the side of the prudent.

Successful build - Chance fights ever on the side of the prudent.

Add hotloading

Successful build - Chance fights ever on the side of the prudent.

Successful build - Chance fights ever on the side of the prudent.

Covert to class to keep cache exclusive to a ScriptInterface

The cache cannot be shared over interfaces as the modules themselves are bound to a compartment.

High-level comments:

• I think this is something we should move towards. It's saner than our current code in the GUI IMO, and could lower load times and sort out the common mess somewhat efficiently.
• I think you'll need to have the module cache at the ScriptInterface level but I would need to check the SM docs
• We have a question for loading whole directories, but maybe we should treat it as an implicit import * as FILENAME from FILENAME.js
• We need to consider how to handle hot reloading as well.

I find this an improvement over the status quo.

Successful build - Chance fights ever on the side of the prudent.

Successful build - Chance fights ever on the side of the prudent.

In D4461#190337, @Langbart wrote:

Its better, but they still switch unitAI gather/approching state.
EDIT1: better video, which compares pushing with rP26244 (left) vs rP26261 (right)
[ Video ]

Commonalize code, maybe fix serialisation issue, am confused by why it seemed to work before.

In D4629#197063, @Freagarach wrote:

You'll need to update cmpUnitMotionFlying to also update the clearance in cmpObstruction.

Successful build - Chance fights ever on the side of the prudent.

Successful build - Chance fights ever on the side of the prudent.

Successful build - Chance fights ever on the side of the prudent.

Successful build - Chance fights ever on the side of the prudent.

remove Civilisations.js, failed to notice that.

• Move gamesettings from gui/gamesettings/ to gamesettings/. This moves attributes as well.
• Move autostart files from gui/autostart/ to autostart/. Note that gui/autostart/ still exists as it's used in the tutorial button from the main menu.
• Implement a placeholder autostart/entrypoint.js in the mod mod.
• Rename the JSI_VFS functions to be explicitly about their function: the ability to write and the potential restricted folders.
• Load the full autostart/ folder instead of just the entrypoint script to allow easier extensions in mods.
• Don't require autostart when using --autostart-client in non-visual mode.

In D4461#197143, @marder wrote:

To me it seems like this is just a bad edge case since the house is so close? I didn't notice that it would happen extensively, but maybe I'm wrong.

To me it seems like this is just a bad edge case since the house is so close? I didn't notice that it would happen extensively, but maybe I'm wrong.

Successful build - Chance fights ever on the side of the prudent.

Successful build - Chance fights ever on the side of the prudent.

spaces

In D4486#197132, @Freagarach wrote:

In D4486#197129, @Stan wrote:

So do you and @wraitii agree on this or are you still debating which of the patches is the right move ?

Good question. :) I haven't seen a formal agreement on the patch level, but our discussion on Friday seemed to hint to this patch.

I think we've largely agreed that long-term, the proper fix here is to implement a templateLoader-level ListTemplates(TemplateType) function, and use different root elements (<Entity>, <Technology>, <Civilization>).

In D4486#197129, @Stan wrote:

Was the message used by the AI?

Nope. Do notice the message was already not sent anymore. I only remove the declaration and the listener in cmpTrainer.

Was the message used by the AI?

It looks fine. I don't notice a performance difference.

Successful build - Chance fights ever on the side of the prudent.

Successful build - Chance fights ever on the side of the prudent.

Purge obsolete message.

See discussion on IRC yesterday.
https://irclogs.wildfiregames.com/%230ad-dev/2022-05-06-QuakeNet-%230ad-dev.log