FOSS gamedev and creative worlds

Today I was building a castle and was scrolling through the inventory and noticed that there are tracks but not carts. Why are carts not in the default game and since they are not why do we have tracks? I think tracks should be tacken out or carts should be put in.

You're not the first one to ask...
IMHO, it's easier to get a mod through the "Mods" submenu then to force carts into upstream.

There's a discussion on mesecons. The problem is just like with carts: people consider bot buggy, heavy and laggy.
The progress, however exist. at least some of us DO want shrunk, light, re-coded version with just basic stuff.

The thing is, there IS a mod called carts_boost. It's smaller and faster.
It's more likely ppl would support it.
Maybe we can discuss it instead of the original carts?

4aiman wrote:The thing is, there IS a mod called carts_boost. It's smaller and faster.
It's more likely ppl would support it.
Maybe we can discuss it instead of the original carts?

Thanks for pointing to this mod. (boost_cart)
It's designed to create less server lag, which can result in some weird situations, where the cart skips a node and goes through a wall.

I would like to have carts in the default game, since there are also default rails.

Carts are likely to be used only by a small bunch of people as they have limited use in survival (too much steel needed for the benefit).

On top of that, the cart mods we have are quite complex and some have bad bugs and exploits.

''just test'' server has a carts mod, it's very goof for me, I didn't experienced any bugs with it

MinetestForFun server has the carts mod. Very often the cart gets stuck 5 blocks above the ground.

@Gael de Sailly
Indeed, sometimes the cart get stuck in "air"... And i use the latest (more stable?) carts mod i've found :/

I tried somes carts mods :
- the more buggy was the default cart mod by Pilzadam,
- after i tried the krock cart mod, but carts still very buggy... but better than Pilzadam version,
- and after i tried the Carbone version, i think it's the less buggy, but still buggy a lot...

For the bug "stuck in air" it's only when the cart rises rapidly...

Boats get stuck in the air too, and are in minetest_game

I still think support for rails and similar constructs ought to be added to the client as such. There's no way to avoid lag completely if the server has to send all changes to the client. The client does know where the rail-nodes are and could in theory move the cart with the attached player along its path.

But what about cheating then, Sokomine?
Or, to be more precise, about bug-users? (Whatever the name for 'em exists in English).
Moving too fast is considered as such.
Digging too fast also.

My point, look at any "mobs" mod:
- client predicts the server's "answer" and destroys the entity when it's killed;
- due to lags/huge "ping" server doesn't know that the entity should be removed yet;
- but it does know that the entity in question is a mob and should attack someone;
- server sends new HP to the client;
- server realizes entity should've been removed;
- confusion.

If a client is gonna predict the movement to the unloaded areas, the server it is connected to may "decide" it "moved to fast" and reset a position.

I'm not against carts or client prediction, but client just mustn't know what server doesn't. E.g., if some area is unloaded, and that unloaded part of a track was cached on the client side, then carts should ignore the rail-like nodes and query for them via force_load or by any other means.

Client side predictionis the only way to reduce lag, except from increasing Internet speeds.

rubenwardy wrote:Client side predictionis the only way to reduce lag, except from increasing Internet speeds.

hmm, what about the thing percieved as lag that sometimes happens in singleplayer? Bad internal wiring?

twoelk wrote:

rubenwardy wrote:Client side predictionis the only way to reduce lag, except from increasing Internet speeds.

hmm, what about the thing percieved as lag that sometimes happens in singleplayer? Bad internal wiring?

Interesting question. The client is able to produce lag by itself when there are many mapblock updates happen. (Removing many nodes at once)

Krock wrote:Interesting question. The client is able to produce lag by itself when there are many mapblock updates happen. (Removing many nodes at once)

That's why proller had split the one and only map lock into several smaller ones in FM's code...
But that didn't do much to the carts.
I wonder how Minecraft does it. I got struck in a wall only once. And that was due to a bad timing of my circuitry (several pistons were too late to open the gate for me).

How about this:
Someone builds a map with lots of rails using minetest_game + some kind of carts.
That certain someone make it so carts behave strangely/lag/whatever.
Someone test that map within MT and FM using the same minetest_game + some kind of carts and running both on the same HW (VBox, anyone?)?

Clientside prediction of what happens on the server and serverside prediction of what clients are going to do has been default for at least two decades.
Maybe have a look at how the quake engine did it.