Minetest_game: Adding 'boost cart' mod to MTGame

[paramat]

An issue for discussing how to add 'boost cart' mod by Krock / SmallJoker.
https://forum.minetest.net/viewtopic.php?f=11&t=10172
I added this to the 0.4.15 milestone but it may not be ready in time.
Addresses part of #515

Experiences with the mod and any issues with it.
What needs rewriting.
Which features (if any) to trim for simplicity and performance.
Whether default rails should be moved into the mod and dealing with backwards compatibility.

@SmallJoker

[paramat]

I guess it should be called 'carts' when added? Or is it better to distinguish it from PilzAdam's mod?

[stujones11]

While I would love to see decent railcarts in minetest, I do not think there is a stable enough implementation yet. boost_cart is a good mod but should not be included with mt game atm (imo)

[Fixer-007]

It is stable enough in my opinion (99% problem free), easy to ride, it behaves much more better than PilzAdam original one. I travelled on them for tens of thousands of blocks on few servers for months. It is stable for production, it was never a problem on servers I played even with big network lag and moderate server one.

I've recommended it for ESM (@maikerumine), Cash World, Vanessa Creative (@VanessaE) and results were good (at least for CW and ESM, much better than original carts I rode on JT, that was horrible experience). @tenplus1 Xanadu server also uses it. I hope they will describe their experience with boost_cart.

I also not buying performance argument, JT 2, xanadu, etc can hold up to 40-50 players, and carts never stalled the server at least when I played with them. You can profile it with mod_profiler if you are unsure about performance. Only few people use carts at one moment on given server, I doubt it uses much cpu. Mobs can use much more CPU, but we are talking about carts used by handful of players. Even pocket calculators can handle that. Boost cart fixes most of PilzAdam carts annoyances I know. I don't know if there any alternatives.

Actual admins who uses it like @maikerumine @tenplus1 @VanessaE should speak up about it more.

[stujones11]

servers are using it without problems

You could say that about many other mods like mob_redo for example, players will love it but expect an issue fest ;-)

[Megaf]

Not happening, carts and mobs are way too unstable in performance. The whole mod thing has to change and how we deal with active objects.

[SmallJoker]

I like the idea to finally give the rails an use. We already have boats and they're part of MTG for a long time.
The performance of the boost_cart is way better than the original version from PilzAdam, compared to the boats it's maybe on the same level. It is proven that servers have no problems with it. The question is, if the singleplayer mode works good aswell.
But - Is it a good idea to extend MTG? Why not add mobs too? We would end up with a huge game that offers all content the players ever will need. I think the players should learn to install mods on their own.

[tenplus1]

Boost_cart works really well on Xanadu with no issues, even added cart sounds and drops as item when cart unattended...

[paramat]

if the singleplayer mode works good aswell

What are the issues in singleplayer?

Is it a good idea to extend MTG? Why not add mobs too?

My opinion is that important and much desired features such as carts and very simple mobs (addressing #515 ) should be moved into MTGame (and be optional in the case of mobs), because any mod within MTGame tends to be far better maintained and kept up to date than an independent mod.

Rails are in default so we really should have carts too, it's embarassing. Boats can be buggy with lua lag and carts will be too but that's no reason to avoid them, who would want to remove boats from MTGame?

[ghost]

boost_cart overrides default:rail, however, and it can act differently depending on the presence of the mod.

I bring this up because if you put this mod on an pre-existing world which features placed rails, then there can be glitches. boost_cart works reasonably well, but I would expect some issues, anyway.

[maikerumine]

I would like to chime in here. I have been hosting very active servers for the past year and a half and have used boost_cart as the main cart mod and have had no problems whatsoever. I feel it would be a neccessary mod to add to the default minetest game. What is the point of having default:rail if you cannot even use them? Moreover, I personally have ridden tens of thousands of blocks with boost cart on fully loaded servers 30+ players and have had no issues. it is time we move forward instead of sitting in the dust, himming and hawing about things that really are not that important.

[stujones11]

I guess it can do no harm to give it go but I think it should be stripped back a little feature wise, the boost_cart mod could always become an add-on pack for those removed. More specifically, all the mesecon dependant stuff needs to be removed and the code style made consistent with existing mtg mods.

[rubenwardy]

It's okay for carts to opt depend on mesecons

[stujones11]

It's okay for carts to opt depend on mesecons

Soft dependency is okay I suppose but in general mtg mods should not depend on outside mods. I guess mesecons is something of an exception so long as items have some other use without it.

[Ferk]

Is it actually necessary for it to depend on mesecons?

It seems the "boost_cart:detectorrail" is the only thing in "boost_cart" that uses mesecons, perhaps that could be stripped out and moved to an external mod. I mean.. that type of rail can't even be used without mesecons anyway, why should vanilla MTG maintain that code? It would also make it harder for mesecons to support old MTG versions with these carts if they ever wanted to change the mesecons API.

The logic that triggers the detectorrail to act could be generalized by using a on_cart_step callback defined in the nodedef of the rail node, if it exists. The boost_cart:is_rail function that is called in the on_step of the cart could give the node back as a second return value, which could be used for this. That would also save a second get_node call.

This way mods could add any other special rail nodes with whatever other effects too. Be it mesecons or whatever other tech.

[sofar]

please review and discuss #1318.

[paramat]

Yeah anything in there that's only usable with mesecons should be removed.

[sofar]

I did remove all of that (bnth moreores and mesecons) entirely in the #1318 PR.

[amadin]

Cart must be activate (drive) by pressing button too, not only by pushing inside or outside cart. First time playng Minetest i long time not undestand how it drive.

[paramat]

Not needed, that's what punch is for anyway, to get the cart rolling onto power rails.

[Wuzzy2]

The idea as such is a real no-brainer. I mean, come on! We have rails but nothing to put on the rails? This just doesn't make sense.

I think boost_cart is a stable and mature mod, I am not aware of any serious issues. It has been used on @test@ for a while now and it works nicely! It even has sounds, something which boats is lacking.
If you know of any bug, please post it here to warn us.

I think the wielding image could look better (it's a bit flat) but that's it.

Note that boost_cart not only includes the carts, but also 3 new rail types.

[amadin]

Not needed, that's what punch is for anyway, to get the cart rolling onto power rails.

Button can't be used as a push button? I mean player must press button again and again for moving in cart on steel rails. So this will work with power rails as usual.

[paramat]

Well, i'm fairly neutral about an external button, but i doubt sofar wants to include this feature, although it seems the API will allow it by registering your own custom rails.

[paramat]

1318 is merged.