bokomatic wrote:As far as I understand floating values exports as 64 bits double (zlib packed and rebased).
Kind-of. The doubles are "cast" to 64-bit ints before being compressed, and back again when uncompressed (as demonstrated in my Java code). Casting to/from 64-bit ints are specified to be lossless operations in both Java and (IIRC) Flash.
May be there is some extended -> double conversion or any other conversion just before export data? Or just after import?
I don't know what happens in the Flash code, and whether it might do anything like this on some platforms. Ryan may have more info.