Actually, Ray, there is a problem with (tblnext), and you happen to be the
fourth person to run across it in the last week and a half. It always seems
to be that way — a problem will exist in the released version for two years,
and suddenly everyone will run into it at once.
Anyhow, AutoLISP is not handling (tblnext)'s second argument properly. If a
NIL value is supplied, it will pick up a value from a random memory location
(which is why it sometimes works and sometimes doesn't). On the 32-bit
workstations, accessing a random memory location is a definite no-no, so this
error causes a "segmentation violation" error on those machines. We're
working on a fix for the next update of AutoLISP — until then, if you don't
want the table rewound, use (tblnext) with just one argument.