Changed usage of sumlink/realpath  in fn_format() to be compatbile
with old MySQL code

commit - it compiles, and test suite still passes, but it is not finished . The
purpose of the commit is backup/code sharing/code review

cleanup of register_slave_on_master()

into mysql.sashanet.com:/home/sasha/src/bk/mysql-4.0

    fixed bad merge

into mysql.sashanet.com:/home/sasha/src/bk/mysql-4.0

into mysql.sashanet.com:/home/sasha/src/bk/mysql-4.0

test case included ( drop) - currently fails, along with rpl000009

error message if the table open or the index creation failed.
Updated portuguese error messages.
Fix for OS/2 that affected CHECK TABLE.

into mysql.sashanet.com:/home/sasha/src/bk/mysql-4.0

Don't force repair of tables that are not closed properly.

Makefile.shared	removed automake complaints about "sqlobjects"
.del-violite.c~53d4251a69d3c	Delete: sql/violite.c

myisamchk: Don't force a repair if the only problem was that the
open count wasn't correct.
Added missing error messages.

into work.mysql.com:/home/bk/mysql-4.0

into mysql.sashanet.com:/home/sasha/src/bk/mysql-4.0

into mysql.sashanet.com:/home/sasha/src/bk/mysql

into hundin.mysql.fi:/home/tonu/mysql-4.0

sync0sync.ic	Had deleted inadvertently lock word = 0 from the non-gcc+i86 case in mutex_reset_lock_word

sync0sync.ic	Use XCHG also to reset the mutex lock word: it makes a serialization point to code on Intel and gives more safety

Added preliminary handling of symlinks in MyISAM.
When using myisamchk to check tables with --force, don't repair
tables that are marked as 'not closed' if they are ok.
Change fn_format() to use my_real_path and my_symlink