numerical error handling in SPSS matrix language

[Ben Pfaff]

I've implemented most of the SPSS matrix language and I'm now working
on leftover pieces such as testing and documentation. One thing that's
come up, that SPSS does not document, is how numerical errors are
handled.  Can anyone tell me what, for example, the following would do
in SPSS?

MATRIX.
PRINT LN(0).
END MATRIX.