into  whalegate.ndb.mysql.com:/home/tomas/mysql-5.1-telco-gca

into  whalegate.ndb.mysql.com:/home/tomas/mysql-5.1-telco-gca

  
The query-cache watch thread was continually allocating new thread entries on the
THD MEM_ROOT, not freed until server exit.
  
Fixed by using a simple array, auto-expanded as necessary.

- do not try to perge binlog when purge comes from slave, it never purges binlog anyways

- also run this patch on 32 bit machines to allow online upgrade

- increase var part reference from 32 to 64 bits on 64 bit systems
- note current patch does not allow for online upgrade from 32 to 64 bit systems or vice versa

- error in initial signal length giving uninitialized variable

into  whalegate.ndb.mysql.com:/home/tomas/mysql-5.1-telco-gca

- make sure all extra flags are reset

into  whalegate.ndb.mysql.com:/home/tomas/mysql-5.1-telco-gca

  - initialize to NULL, to avoid call of free on uninitialized variable

into  whalegate.ndb.mysql.com:/home/tomas/mysql-5.1-telco-gca

into  perch.ndb.mysql.com:/home/jonas/src/51-telco-gca

  make sure master is sendable

  make sure master node is sendable, when getting ref::NotMaster 

- initialize to NULL, to avoid call of free on uninitialized variable

into  perch.ndb.mysql.com:/home/jonas/src/51-telco-gca

    Handle race condtition between MASTER_GCPCONF and execGCP_NODEFINISH

  Fix correct min-value for LockPagesInMemory

  Fix compile error with gcc4

  Fix race in testprg...causing random TC crashes

 increase some timeouts...

  Allow readTablePk to stumble on scan+deleted tuple,
      reporting no-match instead of crash (in case scan is lock-owner)

  Fix bug in SUMA::resend_bucket which could cause mysqld to crash

into  perch.ndb.mysql.com:/home/jonas/src/51-telco-gca

  reset rowcount on temporary error during scan of base table

  fix test_event -n EventOperationApplier

    Handle API failure during resend
    API failure could cause release of table object, which will make resend crash
      when dereferencing table object
      
    Solution, use table_id+hash+schemaversion instead of *raw* pointer in resend

  Make sure not to handle API_FAILREQ if it's already handled

  Make sure head after undo execute does not point to last page of file
  As this will confuse next write to group

  merge to 5.1, adopt testprg to optimized node recovery