Ouch,

modifying rescaled ships and especially **pasted block groups sure is a bad idea**.When rescaling a block group, the group's size will be rounded by the Scale Size: so if you have seven 1x1x1 blocks in a X-row, and scale the group to 10x1x1, the seven blocks will have individual sizes of **10/7** x1x1, so **1,428571...**x1x1.

Multiply that issue with three dimensions and you'll soon get "stranded" on "Match Block" dependency and forced to make imprecise fixes and endure misalignments.There's also the floating point errors, and I also

*hate* them with a

*passion*...

I don't understand how floating point errors even happen with numbers with no decimals, and even powers of two at that !

It seems the contents of the "block info" dialog are loaded into the block size pointer, then reloaded back to the block info, causing the floating point error to appear.

Of course

using numbers who aren't made of additions of powers of two (or fractions thereof) will cause bigger rounding errors due to how floating point variables are stocked (so, for the computer which counts in binary, 0.2 = 0.125 (1/8) + 0.0625 (1/16) + 0.0078125 (1/128) + 0.00390625 (1/256) + 0.00048828125 (1/2048), so happens to be a "real" number with infinite decimals in binary).

There should be a 3-decimal rounding rule to avoid the floating point errors, as 0.999999x1x1 blocks cannot house size-1 turrets and

*aaagh* Right now, with hardpoints, I try rounding to the desired value when putting the block when there's only one erroneous value, and when both hardpoint sizes are erroneous values, I add .01 to those values to overcompensate from the error.