Is it my math or is the physics engine borked?

OK I have mild dyscalculia so it is very possible my math is off here even though I'm using a calculator.

I have a 1st story wall, 8 meters high, with a Z position of 25.750.

The 2nd story wall is supposed to sit immediately on top of it. (The floors are inset.) The 2nd floor wall is 8 meters tall, with a Z position of 33.750.

They should be perfectly stacked, right? But there are visiable seams. It looks like they're off by .005 (you can *just* see a tiny bit of floor flash from the interior floors from some angles) but the coordinates say they are not.

Speaking of .005, I'm also getting vicious prim drift - I smack a prim down at a nice round coordinate like 25.5 and go back a few minutes later to mate another prim to it, and the original prim is now at 24.496 - a number I would never have typed in for any reason.

Everythings acting funny Is it just me or has anyone else noticed this in new builds started since the last upgrade?

I'm prepared to accept that I'm a) crazy or b) cursed, I was just wondering if it's only me

Nopers - yer sane ... My prims are shifting on me as I build lately - visibly shifting - where I see them shift. I will place them back, and poof ... they shift again. Seems as if they have a mind of their own as to what they think will look best ....

Technically this isn't a "Physics" engine problem since I am assuming you aren't enabling "Physics".

Note: The physics engine uses a "Either" collision system where blocks will magically float with a small gap between them.

Also note. Any script with a "Touch" event needs to set "Status_Block_Grab" in order to keep it from being moved a little each time it is touched.

Edit: typos

The prim drifting started getting bad in 1.8 and was never corrected or acknowledged by the lindens. Initially it only affected unlinked prims. Now in 1.9, even linked prim sets drift, especially if you rotate several of them at once.

Its as though they knocked off a few decimal places in precision just to make it less taxing on whatever resources. Who knows.

i dont think its gotten any better or worse, the jumping has gotten abit worse since 1.8
but if you lock your objects down before leaving the region drift is usually minimised

Its gotten worse, and has been bug reported. The bug thats in 1.8 was more related to rotation drift that appeared visually, but not in the edit panel inputs, but can sometimes be corrected by resetting the rotation of all axes to 0 (even though they might already say 0). Locking objects down doesn't do anything. Unless you want to believe that the wind is blowing the prims off alignment.

Hi, Sabrina.

Now that you point it out, I'm seeing sparkles in many of my prim seams. Overlapping them a bit is not really a solution because of texture flashing (and it makes the math harder when placing a lot of prims).

Part of the problem is that the hidden (touching) faces are colored differently by the lighting algorithm, and occasionally get rendered along the seam.

Making those hidden faces transparent helped a little, but then the background sparkles through sometimes. Finally, I stuck a third prim inside of the the two joined ones. Since it was the same color and facing the same way, it didn't matter whether it showed through a bit.

A lot of work just to hide a seam!

Is there a better way, or should I just enjoy the colorful effect?

--Allie

Prim drift has been an issue in SL as long as I can remember. Somewhere in my inventory i have objects that like to drift a few cm each time you edit them (for those I used a script to place them back where they were).

To ensure a position edit the prim and just click inside the edit box for each position number and then click out of the box so it looses focus.

We are all aware of the plain vanilla prim drift. This new form of prim drift is actually harder to correct. I've only been here since 1.6 but there is a noticeable difference.

1.6 and 1.7- Regular prim drift that can be easily corrected as it affected mostly the position as you mentioned above. Most older residents assume that this happens and don't bother to notice anything new beyond this. Loss of naivety perhaps?
1.8 - Rotational prim drift that can be corrected occasionally by resetting all attributes. Happens as a result of rotating and moving more than one prim at once and does not affect rotating linked prims.
1.9 - The same as above, but now affects linked prims. The most worrying 'advancement' in prim drifting.

Forseti mentioned it in this older thread and I have also confirmed this with Torley.

/8/2f/86811/1.html