Re: Optimization techniques and undefined behavior

On 07/05/19 15:03, David Brown wrote:
> If I write a function "square_root" in Go that takes a non-negative
> input and returns its square root, then calling that function with a
> negative input is undefined behaviour.


https://golang.org/pkg/math/#Sqrt


Sqrt returns the square root of x.


Special cases are:


Sqrt(+Inf) = +Inf
Sqrt(±0) = ±0
Sqrt(x < 0) = NaN
Sqrt(NaN) = NaN


Your "square_root" function might have "undefined behaviour" in the
informal sense that the documentation you provide for your function is
incomplete and does not specify what the function returns for a
negative value. But that is not the kind of "undefined behaviour" we
are discussing in this thread: where the *language* says that
"anything could happen" under certain circumstances. Your
implementation of "square_root" will not have "undefined behaviour" in
this technical sense (unless you have perversely implemented a data
race condition!)


There is a reason why the IEEE floating point standard
includes NaN as a result instead of "undefined behaviour".


--
Martin


Dr Martin Ward | Email: martin@gkc.org.uk | http://www.gkc.org.uk
G.K.Chesterton site: http://www.gkc.org.uk/gkc | Erdos number: 4

Return to the comp.compilers page.
Search the comp.compilers archives again.