It looks as if there's a bug in 1.0.0 with FORMAT's dollarsign directive's handling of negative numbers when it has to round up. (Forgive me if this is a known issue--I didn't find it when I searched the bug list and mailing list archives--or if there's something I'm misunderstanding.)
The first group of examples use the F directive for comparison to show that the problem seems to be specific to dollarsign. (I added blank lines between example groups for easy reading.)
The second group of examples illustrates a problem using dollarsign with negative numbers. The output is about 10x too large, and doesn't seem to be getting rounded properly. I'm guessing that the out of bounds error is the result of the same issue.
The third group of examples shows that the problem doesn't seem to occur with positive numbers, although there's still an extra negative sign preprended.
Thank you-
Marshall Abrams
Armed Bear Common Lisp 1.0.0-svn-13663 Java 1.6.0_29 Apple Inc. Java HotSpot(TM) 64-Bit Server VM Low-level initialization completed in 0.7 seconds. Startup completed in 2.211 seconds. Type ":help" for a list of available commands.
CL-USER(1): (format t "~,vf" 3 -0.1768522) -0.177 NIL CL-USER(2): (format t "~,vf" 2 -0.1768522) -0.18 NIL CL-USER(3): (format t "~,vf" 1 -0.1768522) -0.2 NIL CL-USER(4): (format t "~,vf" 0 -0.1768522) -0. NIL
CL-USER(5): (format t "~$" -0.1768522) --1.6 NIL CL-USER(6): (format t "~v$" 3 -0.1768522) --1.75 NIL CL-USER(7): (format t "~v$" 2 -0.1768522) --1.6 NIL CL-USER(8): (format t "~v$" 1 -0.1768522) #<THREAD "interpreter" {10FA1B2D}>: Debugger invoked on condition of type TYPE-ERROR Array index out of bounds: 2 Restarts: 0: TOP-LEVEL Return to top level. [1] CL-USER(9): 0 CL-USER(10): (format t "~v$" 0 -0.1768522) --0. NIL
CL-USER(11): (format t "~$" 0.1768522) 0.18 NIL CL-USER(12): (format t "~v$" 3 0.1768522) 0.177 NIL CL-USER(13): (format t "~v$" 2 0.1768522) 0.18 NIL CL-USER(14): (format t "~v$" 1 0.1768522) 0.2 NIL CL-USER(15): (format t "~v$" 0 0.1768522) 0. NIL