Some work on the missile projectile

[matija-hustic]

An attempt at tackling #7509.

Introduces:

• vertical rate of turn
• customizable vertical launch angle
• delayed activation of the homing mechanism / propulsion
• freefall while propulsion deactivated
• customizable explosion altitude (for airburst)
• customizable cruise altitude (when target out of range)
• height checks for terrain impact and shadow rendering
• acceleration (instead of constant speed from launch to impact)

Not sure about what was meant in #7509 by

• proper 3D math

Still missing and out of scope of this PR (at least currently)

• support for voxel artwork

[reaperrr]

acceleration (instead of constant speed from launch to impact)

Starting with low speed and gradually become faster, rather than flying at constant speed.

[reaperrr]

Might be a good opportunity to unhardcode this. Something like "JammedDiversionBounds" (used for both horizontal and vertical, but there might be a better name for that) with Desc "Defines range of facings by which jammed missiles can stray from current path.".

[ScottNZ]

It's jamming the current path instead of the original path. This gives a compounding jamming effect on the path of the missile.

[reaperrr]

Ok, fixed.

[matija-hustic]

I'll play around with the acceleration next. I guess that that would mean

• an additional upper speed limit will be needed
• too high speeds might adversely affect accuracy
• testing how it'd work with the freefall mode

[GraionDilach]

I don't think Trail and related should be renamed. The tag was named as such in TS/RA2 and theirmods tend to use fire/toxic weapons with them too.

[reaperrr]

Agreed. It's also used for the RA flamethrowers, so it should stay Trail, in my opinion.

[phrohdoh]

So remove the comment?

[GraionDilach]

The current code have them mixed - new tags and the code use Smoke* already with only the old tags aren't renamed - this should be reverted back entirely to use Trail*.

[GraionDilach]

Sorry, I took only a quick glampse so far.

What is freefall mode you refer to? Does that apply to missiles which already passed their RangeLimit but has ExplodeWhenEmpty=false and refers that they fall to the ground without doing(?) damage?

Does ActivationDelay means RA2 CourseLockDuration - missile is forced into the initial facing for x frames and only then starts to guide itself to target?

[GraionDilach]

I haven't tested this yet, true, but unless I overlooked something/didn't grasped the concept, you completely disable acceleration when the firer directly fires upside.

[pchote]

You'll need to rewrite this using integer math. Floating point math can be non-deterministic, and causes hard-to-reproduce desyncs.

[matija-hustic]

This condition might change, but @GraionDilach, right now it works well with any firing angle.

[matija-hustic]

@GraionDilach, you have understood the RangeLimit and ExplodeWhenEmpty == false correctly. But the missile explodes when it touches ground ultimately.

The ActivationDelay also uses freefall (not constant velocity) before the guiding starts. This means that the missile does not just travel in a given direction, but that it follows a ballistic trajectory. You would want to use the LaunchAngle and InitialSpeed here.

[reaperrr]

Nit: Add a 'TODO:' before Rename, easier to keep track of these if you have a buzzword to search for.

[matija-hustic]

OK, updated. The TODO comments will be taken care of when the YAML upgrade rule is in place.

[abcdefg30]

OpenRA.Mods.Common/Effects/Missile.cs(382,36): error CS0103: The name "shadowPos" does not exist in the current context

[matija-hustic]

I've removed the missile shadow part cause it's done in #8923

[matija-hustic]

Making this thing know how to avoid obstacles is proving to be a handful. As it turns out, the previous version of the lookahead didn't handle firing downhill properly...
Now it should finally handle both uphill and downhill firing with some intelligence. But I'll need some more time for testing and going through the code.

@obrakmann, the bike and stealth tank should now be able to fire normally. The previous bug was a consequence of incorrectly choosing negative launch angles.

[reaperrr]

Retested since there have been a few updates since I last looked at this.

The look-ahead works fine now, the vertical angle is factored into the sprite facing and the rest still works. So I'm renewing my 👍

[obrakmann]

Yes, works fine now, 👍

[matija-hustic]

I wanted to clean up the code some more. And make sure no other problems appear... Are you sure you want to merge it in the state it is in?

Can I file a followup PR then? In particular I'm afraid that some of my calls to ISqrt might pass a negative number. And the launch angle & speed selection could be better for downhill firing...

[obrakmann]

Changelog

[obrakmann]

@matija-hustic: oh sorry, I wasn't aware of that. But sure, file a follow-up. It all seemed to work nicely for me, though.

[matija-hustic]

OK, I'll do that asap. And next time I'll ask for a "Has dependencies" label. I just hope the ISqrt calls don't cause problems.

[pchote]

If you do a followup pr, could you please put in some more readable names? We have plenty of spare vowels available for use in variables ;)

[matija-hustic]

Sure :) I would have already left the vowels before were I not afraid of the names being excessively long and ugly. E.g. predAttHght to predictedAttainedHeight basically doubles the length

[GraionDilach]

I'm just here to state that @matija-hustic is a hero:tm:, no more, no less.

[reaperrr]

If you do unhardcode this in a follow-up (I like the idea), personally I'd suggest leaving the minus here and only unhardcode the number, I think that would be more intuitive for modders (higher value = stronger gravity).

[hypercube33]

This seriously junked up the SAM sites for NOD / CnC Gold FYI. The SAMs now shoot missles into the ground short of their targets.