What exactly do current game developers want?

[Senna]

multithreaded engine like old zerver one:) so we can play with 50.000 units and still 100 fps on the new threadrippers

[PicassoCT]

I'm unsure. I think ZK is nearing release so I mostly want nothing to break. I would like it if some time was spent after the 104.0 release to fix lingering issues before rendering is overhauled and needs retesting. Alternately we could have a branch for the bugfixes while the new features are worked on in develop. Both of these requests are more work though.

In a more normal time I would like a focus on polishing core features, such as pathfinding/following. The pathfinding is technically decent but fails to give players a good experience. Godde complained that units would never take the optimal path, they would always oversteer and walk along the 8-directions of the pathfinder, and that he found this frustrating because he would feel forced to give them small move orders just to ensure they take the best path. I made a gadget to solve this (well, heavily modified and enabled a gadget from 2014) but it is a bit unreliable and only works for move commands. Putting it in the engine would be a lot more work as it would need to be compatible with everything. In general, the engine has many good features but there is a lot of work remaining to make them feel great when exposed to players.

I would try to fix that with a widget. In addition to move add channel which allows for two virtual obsticles on the side.

Packages. Made from gameparts. Kept stable by standardized interfaces. Included with a include.

[albator]

player-side, I want MT zerver-mt94-like engine. I upgrade my 9yo cpu and my old i5 750 was still much faster with with mt94 than my current cpu which still has a x2 ipc.

MT is day and night player side

[PicassoCT]

Mostly night on the dev side. As open source endavour spring must uphold code readability sometimes at the price of speed and having a slightly worser performance. Map load is thread pooled. Unless there is a way to do it clean and readable its not going to happen. so noobs lime i can glue funny stuff to the engine.

If zervers implementation would have fulfilled that this part oft the engine would have been the freaj endavour by now.

What do ganedevs want?
Sex, Drugs and Rock n Roll

[Super Mario]

No need for being constructive, be productive instead.

From our last IRC conversation that is so rich of you and hypocritical of you to say that.

In my short presence in this community your net effect has been negative (Almost every thread you make or participate in becomes a flame war).

Stop beating around the bush and just tell me you want me to stop caring and leave this place, I happily obliged.

Consider yourself told.

Indirectly.

[Forboding Angel]

player-side, I want MT zerver-mt94-like engine. I upgrade my 9yo cpu and my old i5 750 was still much faster with with mt94 than my current cpu which still has a x2 ipc.

MT is day and night player side

Small problem with that... Zerver was on a quest to strip out all lua from the engine (picasso alluded to this), which means that you're transported all the way back to 2006. But you can toooooootally have 23049034562345 fleas on the map. #worthit

[bobthedinosaur]

mesh deformation and skeletal support where the bones can be treated like objects as far as scripting goes.

gundams dont have flesh

[PicassoCT]

Smoth switched to the church of neon genesis evangelism.
Bio mechas.
Giant poop.

[PicassoCT]

4) still dreaming of IK from time to time

If that is the case im going to update my build and get it to a usable state.
Be warned though ik is expensive.

Also dont know whether devs will accept a pull request with a new lib by the likes of me. You might have to lend your good name for this endavour.


Also: A "Donate dough for Devhookers" Button on the spring page would be nice

[Silentwings]

Transportation desperately needs to be moved to Lua.

I thought it already was (https://github.com/spring/spring/commit ... 2619f40b30, etc etc) and is just waiting for someone on the game dev side to actually use it.

[AF]

- A half decent CLI interface to Unitsync info
- A nice way of fetching minimaps without needing to mess around with bytes and RGB values, I miss the old unitsync JPEG generator function
- Listing stuff in PR Downloader, the docs are clear on how to grab a game/map but not so much on how to find out what games and maps there are without having a search term

But most of all, decent build instructions on GitHub.

The mediawiki install here is labyrinthine and hard to navigate, and suffers from very poor readability. GitHub markdown is well supported, and would mean visiting the code immediatley gives you instructions on how to build it. It also means if you have a copy of the code you also have all the docs. It also gives us backups in the form of GitHub pages, for which the available themes are far better than anything Mediawiki provides. You can also edit markdown directly in the browser and never have to learn git.

I also believe the wiki should link to these docs rather than duplicating them

[Jools]

Transportation desperately needs to be moved to Lua.

I thought it already was (https://github.com/spring/spring/commit ... 2619f40b30, etc etc) and is just waiting for someone on the game dev side to actually use it.

It's implemented in XTA

[[Fx]Doo]

If possible, i'd like something done about air units:
In UnitDefs table, the TurnRadius value defines 2 things:
- Can a plane directly attack its target, or shall it move away to comeback witha better alignement?
This seems to be the relevant part:
https://github.com/spring/spring/blob/d ... veType.cpp

Code: Select all

	// If goal-position is behind us and goal-distance is less
	// than our turning radius, turn the other way.
	// If goal-distance is half turn radius then turn if
	// goal-position is not in front within a 45 degree arc.
	// This is to prevent becoming stuck in a small circle
	// around goal-position.
	if ((goalDist2D < turnRadius * 0.5f && goalDir2D.dot(frontdir) < 0.7f) || (goalDist2D < turnRadius && goalDir2D.dot(frontdir) < -0.1f)) {
		if (!owner->UnderFirstPersonControl() || owner->fpsControlPlayer->fpsController.mouse2) {
			goalDotRight *= -1.0f;
		}
	}

- Can a plane perform a move order, or is it too close from it, or is it so close that it will already consider having reached its goal?
This seems to be relevant:
https://github.com/spring/spring/blob/d ... AirCAI.cpp

Code: Select all

	// we're either circling or will get to the target in 8 frames
	if ((owner->pos - cmdPos).SqLength2D() < (radius * radius)
			|| (owner->pos + owner->speed*8 - cmdPos).SqLength2D() < 127*127)
	{
		StopMoveAndFinishCommand();
	}

Is it possible to separate those two behaviours with 2 different values in the UnitDef?
Also it seems the second behaviour (move command) uses either TurnRadius*2 when turnRadius > 64 or a 128 default value.
What I have in mind would be these two keys:
attackTurnRadius = ]0;+inf[ (default 500)
moveTurnRadius = ]32; +inf[ 'default attackTurnRadius) and no more using 2x factor in commandAI.

The point of this would be to allow more precise movements while still allowing bombers and other units to perform their attacks correctly.
Now I know the whole point of a large enough TurnRadius was also to avoid planes circling around upon a movecommand that was too close, but it does make managing planes quite *random* and not rewarding enough.

Edit: Got around it via dynamically changing TurnRadius with a gadget depending on wether the next order is an attack order or not. This is not optimal though.

[gajop]

debug.traceback() equivalent for synced Lua.

[gajop]

A way to accurately measure total memory usage

[Jools]

I think you need to elaborate a bit: total memory usage can be monitored by external tools such as Process Explorer.

[gajop]

Right. So I want a way to accurately measure these things (all from within Lua):
- Total Lua memory usage (that counts towards the 800MB memory limit and is displayed in /debug).
- Memory usage of each state separately (it's currently impossible to measure memory usage of synced states afaik).
- Lua/Spring GPU memory usage.
- Total Spring process memory usage (both RAM and GPU memory), and maximum/free system resources.

Most of these commands should be available within unsynced Lua only, with the exception of measuring synced memory usage in synced Lua.

Currently you can only measure how much memory unsynced states are using, and even that you need to synchronize yourself.

[hokomoko]

debug.traceback() equivalent for synced Lua.

hmm, it may be possible to give you a function that echos the trace

[gajop]

hmm, it may be possible to give you a function that echos the trace

Great :D
I'm aware that most, if not all, of these things should be doable, as they're already contained in the engine and displayed/used in various places.

[Kloot]

I didn't write that to ask users for a wish list.

I want

Changed your mind already?

- Total Lua memory usage (that counts towards the 800MB memory limit and is displayed in /debug).
- Memory usage of each state separately (it's currently impossible to measure memory usage of synced states afaik).

Easy and very easy.

- Lua/Spring GPU memory usage.
- Total Spring process memory usage (both RAM and GPU memory), and maximum/free system resources.

Hard (to split them up) and very hard.

should be available within unsynced Lua only, with the exception of measuring synced memory usage in synced Lua

Not possible because things require different amounts of memory on different platforms.