Fixed players smaller than 1 block not being able to walk under overhang

[Thutmose]

With the vanilla code, it is impossible to make a player collide less than 1 block tall, as it will always check the block above for headroom.

This PR adds in a check to see if the player height is less than 1 block, and if so, does not check the block above for headroom.

[Thutmose]

Without this patch, if a mod changes the size of a player to below 1 block tall, the player cannot walk into a space which is only 1 block tall, as it will always check the the block directly above the block containing the player.

This patch allows mods which change the player's size to then let players walk into areas which are 1x1x1.

[simon816]

I think the ideal solution would be to check n blocks above where n is the ceiling of the player's height (applicable to all living entities too).

[Thutmose]

That shouldn't be necessary, the normal collision code (which is run anyway) would cover those cases, as it checks that anyway.

[Thutmose]

Here are some images to explain why this patch is needed:

The hitbox is small enough to fix, but the player cannot walk through

[LexManos]

Actually the PROPER patch would be to remove that second part all together. and potentially allow for some way to detect partially filled blocks.
The Math.round also needs to be converted to a (int)Math.ceil call.
Kinda curious about how @iChun did morph as this is his code. And I know he made it work with partially filled blocks.

[Thutmose]

On the partially filled blocks thing, I was trying to keep it as minimal as possible.

I will test and see what happens if that method is removed entirely (ie always returns true)

[LexManos]

The METHOD doesn't need to be removed, the second block check does because the caller method itterates through all blocks.
Also, this may be worth patching in Entity itself as it ONLY checks 1 block.

[Thutmose]

the call for the method is this: this.pushOutOfBlocks(this.posX - (double)this.width * 0.35D, this.getEntityBoundingBox().minY + 0.5D, this.posZ + (double)this.width * 0.35D);

for each combination of signs for the (double)this.width * 0.35D

If I replace
return !this.worldObj.getBlockState(pos).getBlock().isNormalCube() && (this.height < 1 || !this.worldObj.getBlockState(pos.up()).getBlock().isNormalCube());
with
return !this.worldObj.getBlockState(pos).getBlock().isNormalCube();

Then it will no longer properly push the player out of a block if it just goes over the head, like in this setup:

@simon816 was somewhat correct in saying that ideally it should check up more, given that the initial call is only from the foot location

Edit: The one in Entity only seems to be called by items and xporbs for some reason.

[Thutmose]

Once this section is patched, collision works fine in partial blocks,

and

That particular method is only used for pushing the player out of a block if they manage to get into it in the first place

Should I push the patch with the Math.round turned to (int)Math.ciel?

[LexManos]

Ah I see the issue.
the isHeadspaceFree needs a ! in the if IsOpenBlock call. Once that's done, removing the second block check will work fine.

[Thutmose]

Yep, that seems to do it, re-generating patches with that change.

[iChun]

I never had any problems with players being smaller than one block as long as the eye height of the player was appropriate...

[iChun]

Unless this is a new issue in 1.9, I've not had any issues in MC 1.7.10 or 1.8.

[Thutmose]

did you try in 1.8.9?

I have the player's eye height adjusted correctly (I assume it is, it moves the view point appropriately, the f3+b box shows correctly, and can place blocks directly above myself, and it doesn't suffocate)

It just won't let the player walk into an area with no headroom above the block the player is in.

Here is what I have been using:

https://github.com/Thutmose/Pokecube/blob/master/Pokecube%20Addons/src/main/java/pokecube/pokeplayer/Proxy.java#L163-L189

I guess I should probably use an AT to make setSize(float width, float height) in Entity public.

Edit: as you can see in the above images, it does collide properly, and the eye height seems to be in the correct spot.

[iChun]

No, I've not tested in 1.8.9. I only started development on 1.8.9 a few weeks back and had nothing to do with eyeHeight.

How does elytra in 1.9 handle having the player smaller than a block then? :/

[Thutmose]

I haven't tried my player as pokemob for 1.9 yet, I was getting it written on 1.8.9 first due to the less bugs.

I was going to see about porting it once I got the basics working.

[iChun]

I was gonna say to hold off on this PR till it's been tested in 1.9, then I realised that master is still on 1.8.9. Hmm.

[LexManos]

Ya, nobody seems to change the player's size so nobody has run into this before. However it's a issue between 1.8.9 and 1.9, so it's worth patching both.
Simple fix tho.

[Thutmose]

@iChun I checked 1.7.10, and it doesn't check the block above, which would explain why you never came across this problem there.

I don't have a 1.8 environment setup to check, but it might also be the same there too.

[Thutmose]

So is there anything else I need to do for this?

[LexManos]

Squash, and one for 1.9 would be appreciated {So I can pull both and fix both branches}.

[Thutmose]

I Think I squashed it? I will do one for 1.9.

[LexManos]

You squashed but now it has a useless commit message. Remember, this is what creates the changelog. A "Rebase and recommit" line isn't informative.

[Thutmose]

That should be better description?

[Thutmose]

#2618 was made with the same changes but for 1.9 branch.

The patch file is slightly different due to isNormalCube() being moved to the blockstate instead of in the block

[LexManos]

God dammet was this event tested? Went and tested it in game and had ALL kinda of issues. Like being sucked into the side of walls, and being pushed in the wrong directions.
f14f77d

IChun do me a favor, do 1.8.9's version.