Some of the buggy behaviour is due to state change temperatures being wrong (looks like most of these may be fixed in cracker64's version)
For example:

WTRV freezing point and SNOW melting point were not converted to Kelvin - should be 273.15 K not 0 K.

LN2 freezing and NICE melting temperatures miscalculated - should be approximately 63 K.

O2 condensing and LO2 boiling temperatures are inconsistent with each other, and both are inconsistent with real life value of 90 K.

YEST "melting" to DYST is currently at 273.15 K (hence it is dying as soon as it is created), should be >= 317 K (317 K is currently the max temperature for yeast growth to occur)

Yeast grows from 30-44C and should die at 80C(353K)
I'm pretty sure my version has this and fixes all the other bugs as well
Which is here: https://powdertoy.co.uk/Forum.html?action=vthread&forum=5&topic=863

Problem due to expansion of temperature range:

The temperatures stored in saves (and stamps and copy/paste) are now only accurate to within about 39 degrees.

(temperatures are currently stored as an 8 bit value in saves so precision is ([max temp]-[min temp])/255 = 9999/255 = ~39C)


This causes problems with water, for example - copy some room temperature water, then watch it freeze when you paste it...
(same thing happens in saves done from the beta version - room temperature water freezes on load)

here is a new way to save temperature, it has accuracy to less than a degree, it saves it in 2 8 bit values so temp could be even higher if wanted.
it does not load other saves, because right now it only reads its own saves, but if everyone uses this then there is no problem.
http://www.box.net/shared/i113lc028s

Finalflash50
no

i just finished messing around with some making some magnets, and i'm gonna say NO
to calculate distances between particles on the scale of thousands per frame and moving them the program is laggy and glitchy as hell.
im sure it can be done, but not with this code.
just use black hole to pull things