[armedbear] #190: FORMAT DOLLARSIGN fails to handle rounding negative arguments

armedbear armedbear-devel at common-lisp.net
Sat Nov 16 14:16:18 UTC 2019


#190: FORMAT DOLLARSIGN fails to handle rounding negative arguments
----------------------------------+------------------------------
      Reporter:  Evenson Not Org  |      Owner:  ehuelsmann
          Type:  defect           |     Status:  new
      Priority:  critical         |  Milestone:  1.6.0
     Component:  interpreter      |    Version:  1.0
    Resolution:                   |   Keywords:  ansi-conformance
Parent Tickets:                   |
----------------------------------+------------------------------
Description changed by Evenson Not Org:

Old description:

> In <http://article.gmane.org/gmane.lisp.armedbear.devel/2138> Marshall
> Abrams reports:
>

> 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
> prepended.
>
> Thank you-
>
> Marshall Abrams
>
> * FAILURES
>
> 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
> }}}

New description:

 In <http://article.gmane.org/gmane.lisp.armedbear.devel/2138> Marshall
 Abrams reports:


 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
 prepended.

 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
 }}}

--

--
Ticket URL: <http://abcl.org/trac/ticket/190#comment:14>
armedbear <http://abcl.org>
armedbear


More information about the armedbear-ticket mailing list