Minetest VS Minecraft

[lkjoel]

Yes, exactly. If, persay, it used something like angelscript or pawn, then it would be so much faster, and plus, there wouldn't be as many useless mods

Why would that cut down on the number of useless "mods"?

Because then beginners wouldn't work on mods, and people would think twice before starting one.

[rubenwardy]

I am trying to make it as non-bias as possible.

http://irc.minetest.ru/minetest/2013-03-12#i_2927751

[rubenwardy]

As I said, it's an issue with your configuration. Minetest is written in C++, but has lua mods, and lua is actually really slow. And minetest doesn't use any shaders, which makes it even more slow.

Minetest does have shaders, see the shaders folder in "client"

Lua is only run at server start up time.
minetest.register_<etc> registers a value in a table, so lua files do not slow mt down at all, and is not even run at game loop.

the problem is when functions are registered to run (eg: ATMS) but these are generally fairly fast, except for mods like Mobf and SimpleMobs, because they use heavy weight functions.

Pathfinding and ai will/may be added to core soon (I think sapier/jeija is working on it)

[Dopium]

Well i started watching minecraft videos on you tube and wanted to play it, although i dont really like to pay for games over the internet, so i looked for something else much the same. obviously i found minetest 0.3 as at the stage 0.3 was the latest minetest build and was far behind minecraft. For me minecraft had "greener grass" so to speak and i always wished i had minecraft, however over the last 15 months or so i think minetest has taken off. I played minecraft a few weeks back on a friends PC and was very disappointed, i feel minetest is alot more versital now. There is so much more to do and with certain mods i believe i have a better game then minecraft. No disrespect to minecraft or anything but i think open source is the way to go. The comuntity makes mods the player wants, rather then a company calling all the shots. To me Minetest is already starting to take over minecraft, but thats my personal opinion. 30000 block limit in all directions is alot better then minecrafts capped map. So many things i like more about minetest, i have no interest in minecraft anymore. Only thing i wish is the community will grow a little larger, more people interested in the game would be good. But im happy with the way minetest is coming along

[rubenwardy]

Well i started watching minecraft videos on you tube and wanted to play it, although i dont really like to pay for games over the internet, so i looked for something else much the same. obviously i found minetest 0.3 as at the stage 0.3 was the latest minetest at the time and was far behind minecraft. For me minecraft had "greener grass" so to speak and i always wished i had minecraft, however over the last 15 months or so i think minetest has taken off. I played minecraft a few weeks back on a friends PC and was very disappointed, i feel minetest is alot more versital now. There is so much more to do and with certain mods i believe i have a better game then minecraft. No disrespect to minecraft or anything but i think open source is the way to go. The comuntity makes mods the player wants, rather then a company calling all the shots. To me Minetest is already starting to take over minecraft, but thats my personal opinion. 30000 block limit in all directions is alot better then minecrafts capped map. So many things i like more about minetest, i have no intreast in minecraft anymore. Only thing i wish is the community will grow a little larger, more people interested in the game would be good. But im happy with the way minetest is coming along

Mc's map is only capped vertically to 255.
It extends to 4,000,000 blocks in every horizontal direction, compared to Mt's 31,000.

[Dopium]

Ah ok thanks i didnt look into it but i knew the vertical limit of 256 that i thought was really lame

[lkjoel]

As I said, it's an issue with your configuration. Minetest is written in C++, but has lua mods, and lua is actually really slow. And minetest doesn't use any shaders, which makes it even more slow.

Minetest does have shaders, see the shaders folder in "client"

Lua is only run at server start up time.
minetest.register_<etc> registers a value in a table, so lua files do not slow mt down at all, and is not even run at game loop.

the problem is when functions are registered to run (eg: ATMS) but these are generally fairly fast, except for mods like Mobf and SimpleMobs, because they use heavy weight functions.

Pathfinding and ai will/may be added to core soon (I think sapier/jeija is working on it)

I'm sorry, I meant barely any shaders.

[4aiman]

I'll try to be constructive and, since this a comparison thread, I'll talk about what Minetest is lacking.
Not because Minecraft is better, but because there are some things that can make Minetest better.
Also, whether each of you here personally consider lame or not, things below are meant to improve Minetest and make it more interesting by expanding the survival element.
And the last (but not least): things below should be configurable.

So... You know, there only 2 things which keep my friends from switching to minetest...
And those are...
TA-DA!
Mobs and nether!

While first have been "toyed" with (and there are some achievements, yet not-so-playable 'cause of the toughness of minetest's mobs...) the second seems to be impossible.
Another realm itself is possible and (despite it takes a while) there are floatlands and the nether mods. But teleporters take a player to the realm of content-ignore nodes and he/she waits and waits and waits...

I'm not saying that minetest should have some kind of nether. Absolutely not. But a different realm idea itself is good. And a different realm with it's own mobs is better. And having 2 or more realms is awesome!
Maybe even by using a different game/world.
Again, I'm not saying there are no mapgen mods, but they are too slow...

As for comparison in "beauty", things like plants displacement and some particles emitters (which still are not merged!) can move Minetest towards beauty
Of course all extra graphics should be adjustable. Minetest is suffering from (as someone called this) linux-design disease: there aren't any good artist who can influence coders just enough to let him or her change the default texturepack. Yes, there are lots of things to be worried about rather than that, but there are people who want to help. Yes, new textures won't satisfy all of us, but those should only satisfy more of us than the default ones.

Ok, I know what you want to say: download what you like and - Voilà! But this attitude is the most vicious enemy of the Minetest: there is NO out-of box "prettiness" and "eye-candyness".

So, my point is: alongside with some optimizing, MT needs recommended enhanced "build" of a minetest_game. We already have a minimal_development_test game for tests, so why not make minetest_game better by letting someone who want to make exclusive minetest-only textures for the default game. Also default absolutely must contain at least semi-hostile mobs. The game engine is in development, so any light-weighted enough mobs will do.

[prestidigitator]

...C++ is much faster than Java. Even on a stronger computer, where you can get Java running quickly, C++ is a lower level language, and therefore, is much faster.

This may have been true at one point, but is now a myth. Just-In-Time compiling (JIT), JVM enhancements, and other things have made Java and C/C++ very comparable in terms of performance. In fact, bytecode and JIT sometimes even allow optimizations that are not possible with C/C++ and depending on the hardware, architecture, computing task, and JVM, Java can even outperform C++. See http://en.wikipedia.org/wiki/Java_performance

EDIT: However, I will admit that Java OpenGL implementations still kind of suck at this point. The LWJGL framework that Minecraft uses is one of the best ones as far as I can tell, but more definitely needs to be done in the realm of 3D interfaces.

[Mito551]

Mobs and nether!

what?

nether!

WHAT?
Why? We don't need another realms to my mind. We could have something like different realms in range of 30k blocks. like after 15000 downward there is hell or something and after a thousand upwards floatlands or something followed by somekind of heaven at 15000 or somewhere.

Ok, I know what you want to say: download what you like and - Voilà! But this attitude is the most vicious enemy of the Minetest: there is NO out-of box "prettiness" and "eye-candyness".

So, my point is: alongside with some optimizing, MT needs recommended enhanced "build" of a minetest_game. We already have a minimal_development_test game for tests, so why not make minetest_game better by letting someone who want to make exclusive minetest-only textures for the default game. Also default absolutely must contain at least semi-hostile mobs. The game engine is in development, so any light-weighted enough mobs will do.

ППКС. I totally agree!

[4aiman]

What?
...
WHAT?
...
Why?

I just told everyone about my friends, no need to overreact Different peoples have different likings
(Don't say you never seen anyone to set some Linux shortcuts to be like those of windows or apply the Luna theme

We don't need another realms to my mind. We could have something like different realms in range of 30k blocks. like after 15000 downward there is hell or something and after a thousand upwards floatlands or something followed by somekind of heaven at 15000 or somewhere.

^ In my turn, I can sign below this

Really, that may be a solution - I never stated how this should be implemented
The only thing I stated that Minetest needs some variety.
In fact, if we'll be able to just walk/jump/fall into another realm - it would be even more awesome than different worlds!

[Mito551]

We don't need another realms to my mind. We could have something like different realms in range of 30k blocks. like after 15000 downward there is hell or something and after a thousand upwards floatlands or something followed by somekind of heaven at 15000 or somewhere.

^ In my turn, I can sign below this

Really, that may be a solution - I never stated how this should be implemented
The only thing I stated that Minetest needs some variety.
In fact, if we'll be able to just walk/jump/fall into another realm - it would be even more awesome than different worlds!

Or still make a portal. Like travelnet.

[4aiman]

Or still make a portal. Like travelnet.

Maybe
I like portals, if world around the player will load a bit faster after using them
Also, I vote for an option to disable F5 key functionality - having no idea where I am is more like survival.
(If there's a way to set any other non-existing key instead of F5, then I'll al ears

[Mito551]

just pretend you don't know it, eh?
i dunno, it is really useful in terms of administrating and creativity stuff. you don't want it - you don't use it

[lkjoel]

...C++ is much faster than Java. Even on a stronger computer, where you can get Java running quickly, C++ is a lower level language, and therefore, is much faster.

...
EDIT: However, I will admit that Java OpenGL implementations still kind of suck at this point. The LWJGL framework that Minecraft uses is one of the best ones as far as I can tell, but more definitely needs to be done in the realm of 3D interfaces.

What do you mean?

[prestidigitator]

...C++ is much faster than Java. Even on a stronger computer, where you can get Java running quickly, C++ is a lower level language, and therefore, is much faster.

...
EDIT: However, I will admit that Java OpenGL implementations still kind of suck at this point. The LWJGL framework that Minecraft uses is one of the best ones as far as I can tell, but more definitely needs to be done in the realm of 3D interfaces.

What do you mean?

I mean in terms of raw computation Java does fine, but when interfacing directly with graphics hardware the existing libraries are a bit lacking. They are without exception poorly organized (for example many of them require you to know exactly which version of OpenGL a feature was introduced in to use it, so you have to go searching for every stupid method call you want to make).

The use of direct byte buffers has helped performance quite a bit, but there still needs to be better standard library support so the solution can evolve beyond just wrapping OpenGL libraries in JNI all the time. An example improvement in Linux would be to use memory mapping and the device and sys filesystems to give pure Java interfaces to the graphics hardware without having to drop down to C through JNI (I believe Android does something like this). But since this is very OS/hardware/platform specific, it is unlikely to become a reality until similar methods become feasible in other OSs and more graphics manufacturers sit up and take notice enough to support the effort (e.g. by open-sourcing their drivers).

[lkjoel]

...
EDIT: However, I will admit that Java OpenGL implementations still kind of suck at this point. The LWJGL framework that Minecraft uses is one of the best ones as far as I can tell, but more definitely needs to be done in the realm of 3D interfaces.

What do you mean?

I mean in terms of raw computation Java does fine, but when interfacing directly with graphics hardware the existing libraries are a bit lacking. They are without exception poorly organized (for example many of them require you to know exactly which version of OpenGL a feature was introduced in to use it, so you have to go searching for every stupid method call you want to make).

The use of direct byte buffers has helped performance quite a bit, but there still needs to be better standard library support so the solution can evolve beyond just wrapping OpenGL libraries in JNI all the time. An example improvement in Linux would be to use memory mapping and the device and sys filesystems to give pure Java interfaces to the graphics hardware without having to drop down to C through JNI (I believe Android does something like this). But since this is very OS/hardware/platform specific, it is unlikely to become a reality until similar methods become feasible in other OSs and more graphics manufacturers sit up and take notice enough to support the effort (e.g. by open-sourcing their drivers).

oh I see... That makes sense

[10twenty4]

Regardless of what language is used, Minecraft has some rather shoddy programming contained in it. An easy example is its chunk generation, which is the reason it has such a tiny height limit. Mods exist which make it use cubic chunk generation instead, and suddenly it can support 64000+ layers, much like Minetest does.

[rubenwardy]

Updated the table

http://multa.bugs3.com/minetest/vs/minecraft/

[Calinou]

Supports anaglyphic 3d (red/cyan 3d glasses)

Both support that.

[prestidigitator]

Java has a lot of overhead (slow).

Provably false. See http://en.wikipedia.org/wiki/Java_performance

Java does not work on some computers.

There are probably far more computers Minetest doesn't work on, so this may not be a very good point of comparison.

Java is a inferior game development language.

C++ is an superior game development language.

Both very much opinion not fact, and sound a bit elitist.

It might be better to post benchmarks and supported/known working OS configurations instead of (to be brutally honest) sounding like we just want it to be true. Has anyone tested the maximum reasonable number of players connecting to a vanilla server, for example? If performance is a strength, let's come up with some good numbers we can show.

[lkjoel]

Java has a lot of overhead (slow).

Provably false. See http://en.wikipedia.org/wiki/Java_performance

Java does not work on some computers.

There are probably far more computers Minetest doesn't work on, so this may not be a very good point of comparison.

Java is a inferior game development language.

C++ is an superior game development language.

Both very much opinion not fact, and sound a bit elitist.

It might be better to post benchmarks and supported/known working OS configurations instead of (to be brutally honest) sounding like we just want it to be true. Has anyone tested the maximum reasonable number of players connecting to a vanilla server, for example? If performance is a strength, let's come up with some good numbers we can show.

+100, that's totally true. Java is not an inferior language, it's just a language where you can spend more time working on the actual game instead of the little wee details, plus you can forget about cross-compiling, all the testing is faster, etc... So no, it isn't inferior for game development at all.

[prestidigitator]

Other interesting benchmarks might be for world loading/generation. Can we create measurements for the rate at which chunks are generated and transmitted all the way to the client for a brand new world? How about the rate at which the chunks can be loaded from the database and transmitted for pre-generated portions of an existing world? This could be run on vanilla and with various mods that affect map generation to give a range of neat metrics to show off.

[davetube24]

+1 on mobs with a peaceful mode!

[miapia2331]

]its your choice if u like minetest or minecraft your appinion people mite think your wrong who cares thats what u think so why u guys have a debate whats better