fixes to get flush test to work only to discover another bug in flush

tables - read comments below


client/mysqltest.c:
  fixed bug in send/reap 
  added comments explaining how things work 
  to remove from Monty the temptation to "clean up" my code
  use a separate pointer for the test line buffer from the query itself - in the
  case of send they are not the same
  added memory clean-up for register variables
mysql-test/mysql-test-run.sh:
  added --skip-* option - you can now skip anything you want, good thing
  to skip is bdb and innobase if you are running only one test to speed
  up server startup
mysql-test/t/flush.test:
  fixed up flush test after fixing send/reap bug - now found a new bug
  with flush - I get 1034 incorrect key file error on this - will 
  investigate why, but thought I'd commit the test case first
  
  Another case to support Sinisa's assertion that FLUSH TABLES is dangerous
  and should be avoided :-)
BitKeeper/etc/logging_ok:
  Logging to logging@openlogging.org accepted

1 files changed, 19 insertions, 6 deletions

diff --git a/mysql-test/t/flush.test b/mysql-test/t/flush.test
index 750ecdeadda..5738f217d01 100644
--- a/mysql-test/t/flush.test
+++ b/mysql-test/t/flush.test
@@ -2,11 +2,24 @@ connect (con1,localhost,root,,test,0,mysql-master.sock);
 connect (con2,localhost,root,,test,0,mysql-master.sock);
 connection con1;
 drop table if exists t1;
-create temporary table t1(n int);
-connection con2;
-#send flush tables;
+create temporary table t1(n int not null primary key);
+drop table if exists t2;
+create table t2(n int);
+insert into t2 values(3);
+let $1=100;
+while ($1)
+{
+ connection con1;
+ send replace into t1 select n from t2;
+ connection con2;
+ send flush tables;
+ connection con1;
+ reap;
+ connection con2;
+ reap;
+ dec $1;
+}
+
 connection con1;
-insert into t1 values(3);
 select * from t1;
-connection con2;
-#reap;
+drop table t2;