Tk Library Source Code

View Ticket
Login
Ticket UUID: 1566488
Title: failed tests with Tcl 8.5 (64-bit)
Type: Bug Version: None
Submitter: dgp Created on: 2006-09-27 17:01:36
Subsystem: math :: bignum :: float Assigned To: dgp
Priority: 5 Medium Severity:
Status: Closed Last Modified: 2006-09-29 03:13:09
Resolution: Invalid Closed By: dgp
    Closed on: 2006-09-28 20:13:09
Description:
Using Tcl HEAD (8.5a5) on a
64-bit system:

==== bigfloat-fromdouble-2.1f Some floating-point
computations related to command fromdouble FAILED
==== Contents of test case:
return $resultat
---- Result was:
0
---- Result should have been (exact matching):
1
==== bigfloat-fromdouble-2.1f FAILED


==== bigfloat-int2float-2.9.2.3 Some integer
computations related to command int2float FAILED
==== Contents of test case:
uplevel 1 $code
---- Result was:
-29127.3114
---- Result should have been (exact matching):
5000000000.0000
==== bigfloat-int2float-2.9.2.3 FAILED


==== bigfloat-cos-4.4.1.0 Some floating-point
computations related to command cos FAILED
==== Contents of test case:
return $resultat
---- Result was:
0
---- Result should have been (exact matching):
1
==== bigfloat-cos-4.4.1.0 FAILED

---- bigfloat-sin-4.4.1.1 start


==== bigfloat-sin-4.4.1.1 Some floating-point
computations related to command sin FAILED
==== Contents of test case:
return $resultat
---- Result was:
0
---- Result should have been (exact matching):
1
==== bigfloat-sin-4.4.1.1 FAILED

---- bigfloat-cos-4.4.2.0 start


==== bigfloat-cos-4.4.2.0 Some floating-point
computations related to command cos FAILED
==== Contents of test case:
return $resultat
---- Result was:
0
---- Result should have been (exact matching):
1
==== bigfloat-cos-4.4.2.0 FAILED

---- bigfloat-sin-4.4.2.1 start


==== bigfloat-sin-4.4.2.1 Some floating-point
computations related to command sin FAILED
==== Contents of test case:
return $resultat
---- Result was:
0
---- Result should have been (exact matching):
1
==== bigfloat-sin-4.4.2.1 FAILED
User Comments: dgp added on 2006-09-29 03:13:09:
Logged In: YES 
user_id=80530


Fixing Tcl bug 1567222
stopped all six of
these failures.

dgp added on 2006-09-29 01:33:42:
Logged In: YES 
user_id=80530


At least that one is really
a Tcl bug.  1567222

dgp added on 2006-09-29 01:08:14:
Logged In: YES 
user_id=80530


Selecting bigfloat-int2float-2.9.2.3
for more analysis, on 32-bit system
I see:

% int2float [fromstr 5000000000] 5
F 21474836480000000000 -32 42951

On 64-bit I see:

% int2float [fromstr 5000000000] 5
F -125100849711610 -32 42951

sarnold75 added on 2006-09-28 22:30:06:
Logged In: YES 
user_id=1160227

I am sorry, I have tested the package 2.0 with Tcl HEAD
(updated and recompiled today) Windows XP (32-bit), and
there is no problem - the test suite runs with no error.

I guess the bug is somewhere in 64-bit integers handling,
and/or even bignums.
Maybe others areas of tcllib are impacted, this may be a
problem related to the core, and which does not happen on
32-bt systems.

I feel sorry for you, but I'd only seek for a solution if I
have a 64-bit system under my (dirty :^) hands...