HP PRIME not raising overflow/underflow may cause critical errors

[robve]

A simple textbook example: combinations, aka. nCr binomial coefficients:

Define nCr
X!/Y!*(X-Y)!

Now, this is NOT the most efficient way to compute nCr, but let's put that aside for now (I will comment on that later). We are just defining a classic textbook version of nCr, because we just want to try and apply a function we learned about in a textbook. Sadly, without exceptions being raised, we get garbage:

nCr(270,1)
1

At least an error message would have told us that we better not use this function for large n producing problematic results with low accuracy and overflow.

I am not being facetious. This can be critical. I am aware that most people will not write nCr this way, since we know it is not the most efficient way to compute nCr. The built-in COMB works fine, for example. We can also write some HPPL code to compute nCr the right way. The problem is that user-defined functions and HPPL code cannot be trusted when exceptions aren't raised when you expect them.

- Rob