So I've heard of single structures going up to 20,000 blocks..... Would 150,000 block structures be impossible?

I don't think this would be impossible but before you plan to make this, I do want to say that structures as big as those would induce a lot of lag to the player's client and could crash it, maybe try making more, just smaller scaled.

If you generate it progressively instead of entirely, size won't matter.

@Gotolink: For such massive structures, would using ExtendedBlockStorage to set he blockID's and meta's be better in order to prevent lags while it is being generated?

@Mazetar

If you have organized the generation in chunks, it is a good idea, indeed.

Don't expect much improvement though.

@Mazetar

If you have organized the generation in chunks, it is a good idea, indeed.

Don't expect much improvement though.

 

I know how to reduce lag a fair bit.

 

I had a friend who built a testing structure that used world.setBlock ( args ) for every block he wanted. He had a solid rectangle of dimensions 10 x 10 x 60.

My game lagged like crazy when I generated it.

 

I used a nested for loop to do the exact same structure. I didn't lag at all.

 

So my piece of advice, use as many for loops as you can... Unless it absolutely pointless to do so.

I used a nested for loop to do the exact same structure. I didn't lag at all.

 

So my piece of advice, use as many for loops as you can... Unless it absolutely pointless to do so.

 

I really don't see the logic here

im more concerned at how he did it WITHOUT loops ....

im more concerned at how he did it WITHOUT loops ....

 

Schematic to Java converter makes a class file full of stuff like thos:

 

world.setblock(x+1,y,z, 1)

world.setblock(x+2,y,z, 1)

etc.

 

but I can't imagine the computation of the addition would be the difference between an noticeable amount of lag for 6000 blocks?

im more concerned at how he did it WITHOUT loops ....

 

Schematic to Java converter makes a class file full of stuff like thos:

 

world.setblock(x+1,y,z, 1)

world.setblock(x+2,y,z, 1)

etc.

 

but I can't imagine the computation of the addition would be the difference between an noticeable amount of lag for 6000 blocks?

 

Hell... It lagged for a couple seconds! I don't have a top of the range computer, but it is still good. And yes it lagged. I redid it with the for loops and it didn't lag a bit. hell... My computer lagged just READING the code for it.

 

That convertor SUCKS anyway... The amount of editing you have to do to the files after it converts is horrendous... You have to edit the name of every method, you have to CALL every method.... Oh yeah, and you also get an error in every method because there is too much information in it. So you have to cut ALL the methods into a third of their original size. All in all, I could do an okay structure in one third of the time it would take to build it, convert to a schematic, convert to code, fix the code, and run it.

 

 

I used a nested for loop to do the exact same structure. I didn't lag at all.

 

So my piece of advice, use as many for loops as you can... Unless it absolutely pointless to do so.

 

I really don't see the logic here

 

The logic? I failed that class in school. But what I meant was. Say I wanted to place one block. Don't use a for loop. Say I wanted to place a wall 3 x 3 x 10. Use a for loop.

No with logic I meant, I fail to see any logical reason as to why it would have a significant impact on performance

Hell... It lagged for a couple seconds! I don't have a top of the range computer, but it is still good. And yes it lagged. I redid it with the for loops and it didn't lag a bit. hell... My computer lagged just READING the code for it.

 

That convertor SUCKS anyway... The amount of editing you have to do to the files after it converts is horrendous... You have to edit the name of every method, you have to CALL every method.... Oh yeah, and you also get an error in every method because there is too much information in it. So you have to cut ALL the methods into a third of their original size. All in all, I could do an okay structure in one third of the time it would take to build it, convert to a schematic, convert to code, fix the code, and run it.

 

yeah but if i was doing a converter i would do that too

btw the method too big is a java thing, and it make sens that the converter doesnt check for that, they probably made it for 5x5x5 and less structure so yeah if you have a 1000x1000x1000 you might/will get bytecode too big exception

 

about the code having errors, probably because its outdated, kinda like techne having error when you try to import it

No with logic I meant, I fail to see any logical reason as to why it would have a significant impact on performance

Don't assume any logic behind it, benchmarking on Minecraft terrain generation is useless.

Mew probably felt like it had an impact, while any other change could have done that.

 

Obviously you need to iterate over the blocks, it would be utterly stupid not to.

I prefer to use Iterator by the way