Bug #40180 Test main.xml failing on Windows 'classic' builds
Submitted: 20 Oct 2008 15:31 Modified: 20 Oct 2008 16:37
Reporter: Patrick Crews Email Updates:
Status: Duplicate Impact on me:
None 
Category:Tests: Server Severity:S3 (Non-critical)
Version:5.1 OS:Windows
Assigned to: Assigned Account CPU Architecture:Any
Tags: pushbuild, test failure, windows, XML

[20 Oct 2008 15:31] Patrick Crews
Description:
Test main.xml is failing on Windows classic builds:

19357: main.xml                       [ fail ]
19358: 
19359: mysqltest: At line 549: query 'CREATE TABLE t1(a INT NOT NULL)' failed: 1050: Table 't1' already exists
19360: 
19361: The result from queries just before the failure was:
19362: < snip >
19363: Warning	1292	Truncated incorrect INTEGER value: 'string          '
19364: Warning	1292	Truncated incorrect INTEGER value: 'string          '
19365: DROP PROCEDURE spxml;
19366: select UpdateXML('<a>a</a>',repeat('a b ',1000),'');
19367: ERROR HY000: XPATH syntax error: 'b a b a b a b a b a b a b a b a '
19368: select ExtractValue('<a>a</a>', '/a[@x=@y0123456789_0123456789_0123456789_0123456789]');
19369: ERROR HY000: XPATH error: comparison of two nodesets is not supported: '=@y0123456789_0123456789_0123456'
19370: select ExtractValue('<a>a</a>', '/a[@x=$y0123456789_0123456789_0123456789_0123456789]');
19371: ERROR HY000: Unknown XPATH variable at: '$y0123456789_0123456789_01234567'
19372: select updatexml(NULL, 1, 1), updatexml(1, NULL, 1), updatexml(1, 1, NULL);
19373: updatexml(NULL, 1, 1)	updatexml(1, NULL, 1)	updatexml(1, 1, NULL)
19374: NULL	NULL	NULL
19375: select updatexml(NULL, NULL, 1), updatexml(1, NULL, NULL), 
19376: updatexml(NULL, 1, NULL);
19377: updatexml(NULL, NULL, 1)	updatexml(1, NULL, NULL)	updatexml(NULL, 1, NULL)
19378: NULL	NULL	NULL
19379: select updatexml(NULL, NULL, NULL);
19380: updatexml(NULL, NULL, NULL)
19381: NULL
19382: CREATE TABLE t1(a INT NOT NULL);
1

How to repeat:
On a Windows system, run the test from the mysql-test directory:

./mysql-test-run.pl --force xml

And observe the test results.

Suggested fix:
Fix the test.

Determine if this is a case of bad cleanup, etc.  The fact that this appears to be limited to Windows classic builds suggests that if it is the test, it is an OS-specific hole in the test.
[20 Oct 2008 16:37] Joerg Bruehe
Duplicate of bug#37988 which was declared "Not a bug", as the fault is with the preceding test that didn't cleanup table "t1".

There was mail on the QA team mailing list about this.

The cause is test "windows" (that's why this is OS-specific), it was reported as bug#39958 whose patch is now in the bugteam's team trees.