Bug #7639 Test failure: 'type_bit': 'concat()' seems to yield 0
Submitted: 3 Jan 2005 16:43 Modified: 19 Jan 2005 11:53
Reporter: Joerg Bruehe Email Updates:
Status: Closed Impact on me:
None 
Category:MySQL Server Severity:S3 (Non-critical)
Version:5.0.3-pre OS:Linux (Linux/s390)
Assigned to: Ramil Kalimullin CPU Architecture:Any

[3 Jan 2005 16:43] Joerg Bruehe
Description:
Test 'type_bit' fails in a pull of Dec 31:

-------------------------------------------------------
*** r/type_bit.result   Fri Dec 31 20:00:11 2004
--- r/type_bit.reject   Sat Jan  1 11:18:40 2005
***************
*** 357,365 ****
  update t1 set b= concat(a);
  select a+0, b+0 from t1;
  a+0   b+0
! 1     1
! 3     3
! 2     2
  drop table t1;
  create table t1 (a bit(7), key(a));
  insert into t1 values (44), (57);
--- 357,365 ----
  update t1 set b= concat(a);
  select a+0, b+0 from t1;
  a+0   b+0
! 1     0
! 3     0
! 2     0
  drop table t1;
  create table t1 (a bit(7), key(a));
  insert into t1 values (44), (57);
-------------------------------------------------------

How to repeat:
Run the test suite on 'ds20'.
[3 Jan 2005 16:45] Joerg Bruehe
Typing error:  Occurs on Linux/390  = 'etpglb0'.
[17 Jan 2005 10:43] Joerg Bruehe
The bug has spread onto other platforms: In the current build, based on
ChangeSet
  1.1771 05/01/13 15:10:12 lenz@mysql.com +1 -0
  - keep 5.0.3-alpha version string (after-merge fix)

the same failure occurs on
aix52 (both 32 and 64 bit), bitch, butch (32 + 64), etpglb0 (as original), hp3750 (32 + 64), hpita, hpux (32 + 64), octane2 (32 + 64), pegasos1, powermacg4, powermacg5, sunfire100b (32 + 64), and sunfire100c (32 + 64).

The test passes on barney, bsd53, build, buildqnx, cane, cane-linuxthreads, ds9, hammer, intelxeon, ita2, nocona (icc), quadita2, sol9x86, and production (during Bootstrap).
[19 Jan 2005 11:53] Ramil Kalimullin
fixed in 5.0.3