Bug #16898 Problems with character set and table file names

Description:
Release build of 4.1.18, based on ChangeSet
  1.2463 06/01/27 13:38:15 kent@mysql.com +1 -0
  disabled.def:
    Enabled rpl_* tests now when bug fixed

This symptom is specific to OS X 10.4 (PowerPC), 32 bit, "debug" build.

(I hope the characters do not get changed by copy-paste)

=====
mysqldump                      [ fail ]

Errors are (from PATH/mysqltest-time) :
mysqldump: Couldn't find table: "T_1"

mysqldump: Couldn't find table: "T%1"

mysqldump: Couldn't find table: "T'1"

mysqldump: Couldn't find table: "T_1"

mysqldump: Couldn't find table: "T_"

mysqltest: At line NNN: Result length mismatch
(the last lines may be the most important ones)
Below are the diffs between actual and expected results:
-------------------------------------------------------
*** r/mysqldump.result
--- r/mysqldump.reject
***************
*** 163,169

  /*!40000 ALTER TABLE `t1` DISABLE KEYS */;
  LOCK TABLES `t1` WRITE;
! INSERT INTO `t1` VALUES ('абÑ<86>де');
  INSERT INTO `t1` VALUES (NULL);
  UNLOCK TABLES;
  /*!40000 ALTER TABLE `t1` ENABLE KEYS */;
--- 163,169

  /*!40000 ALTER TABLE `t1` DISABLE KEYS */;
  LOCK TABLES `t1` WRITE;
! INSERT INTO `t1` VALUES ('?????');
  INSERT INTO `t1` VALUES (NULL);
  UNLOCK TABLES;
  /*!40000 ALTER TABLE `t1` ENABLE KEYS */;
***************
*** 417,423

  /*!40000 ALTER TABLE `t1` DISABLE KEYS */;
  LOCK TABLES `t1` WRITE;
! INSERT INTO `t1` VALUES ('Ã<84>Ã<96>Ã<9c>Ã<9f>');
  UNLOCK TABLES;
  /*!40000 ALTER TABLE `t1` ENABLE KEYS */;

--- 417,423

  /*!40000 ALTER TABLE `t1` DISABLE KEYS */;
  LOCK TABLES `t1` WRITE;
! INSERT INTO `t1` VALUES ('ÄÖÜß');
  UNLOCK TABLES;
  /*!40000 ALTER TABLE `t1` ENABLE KEYS */;

***************
*** 441,447

  /*!40000 ALTER TABLE `t1` DISABLE KEYS */;
  LOCK TABLES `t1` WRITE;
! INSERT INTO `t1` VALUES ('<8e><99><9a>á');
  UNLOCK TABLES;
  /*!40000 ALTER TABLE `t1` ENABLE KEYS */;

--- 441,447

  /*!40000 ALTER TABLE `t1` DISABLE KEYS */;
  LOCK TABLES `t1` WRITE;
! INSERT INTO `t1` VALUES ('ÄÖÜß');
  UNLOCK TABLES;
  /*!40000 ALTER TABLE `t1` ENABLE KEYS */;

***************
*** 462,468

  /*!40000 ALTER TABLE `t1` DISABLE KEYS */;
  LOCK TABLES `t1` WRITE;
! INSERT INTO `t1` VALUES ('<8e><99><9a>á');
  UNLOCK TABLES;
  /*!40000 ALTER TABLE `t1` ENABLE KEYS */;

--- 462,468

  /*!40000 ALTER TABLE `t1` DISABLE KEYS */;
  LOCK TABLES `t1` WRITE;
! INSERT INTO `t1` VALUES ('ÄÖÜß');
  UNLOCK TABLES;
  /*!40000 ALTER TABLE `t1` ENABLE KEYS */;

***************
*** 483,489

  /*!40000 ALTER TABLE `t1` DISABLE KEYS */;
  LOCK TABLES `t1` WRITE;
! INSERT INTO `t1` VALUES ('Ã<84>Ã<96>Ã<9c>Ã<9f>');
  UNLOCK TABLES;
  /*!40000 ALTER TABLE `t1` ENABLE KEYS */;

--- 483,489

  /*!40000 ALTER TABLE `t1` DISABLE KEYS */;
  LOCK TABLES `t1` WRITE;
! INSERT INTO `t1` VALUES ('ÄÖÜß');
  UNLOCK TABLES;
  /*!40000 ALTER TABLE `t1` ENABLE KEYS */;

-------------------------------------------------------

How to repeat:
Detected by the test suite.

I could not repeat this problem on powermacg5.

Looking into the diff in the report, I can tell that the "mysqltest.result" part looks incorrect,
while "mysqltest.reject" is correct. It looks like "mysqltest.result" was corrupted
for some reasons when the test was running.

Server part works fine, there's nothing to fix from my side.

Setting status to "can't repeat" and assigning to Joerg.