Another issue in JLS8 section 16.2.15

[Hiram Hunt]

Hello,
     I think I see another issue in JLS8 section 16.2.15.
The phrase

     – V is definitely unassigned after every assert
       statement that occurs in the try block.

occurs in two locations in 16.2.15 as part of a determination
of whether V is definitely unassigned before a later block.
The first case is for before catch blocks and the second case
is for before the finally block.  The problem is that even
if V is definitely unassigned after the assert, V could
still be assigned if the assert completes abruptly and this
execution path will lead to a catch block (maybe) and
to any finally block.  The assignment could occur in either
expression of the two expression form of assert.

     I have not carefully tried every combination of one and
two expression asserts and try, catch and finally blocks, but
javac did prevent double assignment in the cases I checked.

-- Hiram Hunt (hiramhunt at verizon.net)