LB5.353 float underflow vs. LB4x

[tsh73]

The point is, there is no one in LB5.
I dunno, may be it is a good thing.
on error goto [q]
x=1
while x>0
oldx=x
x=x/10
wend
print "no err"
print oldx
end

[q]
print "err", Err$
print oldx

LB4x:
err           Float underflow exception
1.0e-307

LB5.353:
no err
9.8813129168249e-324
It just decreases until it becames 0, no error in a process. (actually sounds good for me)

I really expected it to end at around 1e-308, but no.
Wikipedia::Double-precision_floating-point_format says

The 11 bit width of the exponent allows the representation of numbers between 10−308 and 10308, with full 15–17 decimal digits precision. By compromising precision, the subnormal representation allows even smaller values up to about 5 × 10−324.

So obviously this is "subnormal representation".
Is there something bad to expect out of it?

[Carl Gundel]

Well, it's interesting. Both LB4 and LB5 rely on the floating point implementation of the Smalltalk they are built upon. These two Smalltalk's approach this issue differently.