Bug#433592: clisp: Float/rational comparison sometimes underflows; should never

2008-07-09 Thread Sam Steingold

this is not debian-related.
this is an upstream bug
https://sourceforge.net/tracker/index.php?func=detailaid=2014262group_id=1355atid=101355
which has now been fixed in the CVS.
it will be available in clisp 2.47
in the future, please report such bugs directly to us using the SF bug 
tracker.

thanks!



--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]



Bug#433592: clisp: Float/rational comparison sometimes underflows; should never

2008-07-09 Thread Luca Capello
tags 433592 + upstream
forwarded 433592 
https://sourceforge.net/tracker/index.php?func=detailaid=2014262group_id=1355atid=101355
tags 433592 + fixed-upstream
thanks

Hi Sam!

On Wed, 09 Jul 2008 17:30:36 +0200, Sam Steingold wrote:
 this is not debian-related.
 this is an upstream bug
 https://sourceforge.net/tracker/index.php?func=detailaid=2014262group_id=1355atid=101355
 which has now been fixed in the CVS.
 it will be available in clisp 2.47

Thank you for the information!  I updated the Debian BTS record as well.

 in the future, please report such bugs directly to us using the SF bug 
 tracker.

No, please.  While other Debian maintainers prefer this workflow, I'm
against: when a bug is discovered in Debian, it should be filed in the
Debian BTS.  Then, the Debian maintainer should take care of filing it
upstream.

Thx, bye,
Gismo / Luca


pgpU9CFEXJQKu.pgp
Description: PGP signature


Bug#433592: clisp: Float/rational comparison sometimes underflows; should never

2007-07-17 Thread J.P. Larocque
Package: clisp
Version: 1:2.41-1
Severity: normal

Hi,

I'm trying to compare a floating-point number and a rational.
According to the rule of float and rational contagion, described in
CLHS 12.1.4.1, and X3J13 issue
CONTAGION-ON-NUMERICAL-COMPARISONS:TRANSITIVE, such a comparison
should always return the mathematically correct result, as if the
floating-point number were first converted to a rational with
RATIONAL.  Instead, CLISP signals an underflow condition:

$ clisp -q -x '( (expt 10 -100) least-positive-short-float)'
*** - floating point underflow

The Implementation Notes for GNU CLISP, sections 12.2.3.1 and
12.2.3.2, don't seem to say anything about comparisons, but mention
two variables which control ANSI-compliance.  CLISP behaves the same
when invoked with -ansi, to set these and any similar variables to
T.

$ clisp -ansi -q -x '( (expt 10 -100) least-positive-short-float)'
*** - floating point underflow

This X3J13 issue is listed as supported in the Implementation Notes for GNU 
CLISP:
http://clisp.cons.org/impnotes.html#issues

CLISP behaves as expected if the floating-point number is filtered
through RATIONAL.  (Note that CLHS 12.1.4.1 explicitly states When
rationals and floats are compared by a numerical function, the
function rational is effectively called to convert the float to a
rational and then an exact comparison is performed.)

$ clisp -q -x '( (expt 10 -100) (rational least-positive-short-float))'
T
$ clisp -q -x '( (expt 10 -10) (rational least-positive-short-float))'
NIL

I'm fairly certain CLISP is in error.

Thanks,

-- System Information:
Debian Release: 4.0
  APT prefers stable
  APT policy: (500, 'stable')
Architecture: amd64 (x86_64)
Shell:  /bin/sh linked to /bin/bash
Kernel: Linux 2.6.18-12-custom-xen-ws-1
Locale: LANG=C, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8)

Versions of packages clisp depends on:
ii  common-lisp-controller  6.9  This is a Common Lisp source and c
ii  libc6   2.3.6.ds1-13 GNU C Library: Shared libraries
ii  libice6 1:1.0.1-2X11 Inter-Client Exchange library
ii  libncurses5 5.5-5Shared libraries for terminal hand
ii  libreadline55.2-2GNU readline and history libraries
ii  libsm6  1:1.0.1-3X11 Session Management library
ii  libx11-62:1.0.3-7X11 client-side library
ii  libxext61:1.0.1-2X11 miscellaneous extension librar
ii  libxpm4 1:3.5.5-2X11 pixmap library

clisp recommends no packages.

-- no debconf information

-- 
J.P. Larocque: [EMAIL PROTECTED], [EMAIL PROTECTED]


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]