problems with comparisons between floating point

[Roberto Porcù]

Dear all,

I'm having some troubles with comparisons between floating point numbers:

in my code, the test   

s(end) == tspan(counter)

sometimes fails while it should not fail because the two numbers are really equal.

So I tried to replace it with the comparison:

abs(s(end)-tspan(counter)) / (max(abs(s(end)),abs(tspan(counter)))) < 8*eps

but  this sometimes fails too, because the number 8 seems to be not adequate.

Could anybody give me a help?

Thanks in advance

Roberto P.