This test case involves the "currency character" mentioned in the spec.
As I read the spec (12.3, first paragraph), this pattern should throw
an error, but nobody does. They just treat the character as they do
all other non-special characters, making part of the fixed prefix string.
Anyone want to argue for a different behavior?
Reminder: I have yet to see a spec, from the JDK or elsewhere, that has
sufficient detail to explain all the nuances of formatting per the
format string. In the case at hand, we should note that the method of
determining where the prefix stops and the special-character part
begins is fuzzy. Assuming # and . have their default meanings,
in "FOO###.#" the prefix is "FOO";
in "F¤OO###.#" the prefix is "F¤OO";
in "FOO¤###.#" the prefix is "FOO¤" according to today's
processors, but "¤###.#" should not yield a ¤ nor a $
prefix. In a strict reading of the spec, any occurrence of ¤
anywhere in the format string is grounds for raising an error.
.................David Marston
