We need to perform testing wrt the following items:
[1] Large no. of Inserts(10,000 tuples).
[2] Test for Table with unique columns too.
[3] Test the error handling capabilites of the engine & parser.
[4] Test for speed with autocommit enabled & disabled, and compare the 2 speeds.
[5] Repeated SELECT of a large table.
[6] Same with JOINs.
[7] [5] & [6] with complex JOIN conditions.
[8] Check the speed of selects. This should be done to check if the
page cache is having any speedup influence. Generally 1st select should
be the slowest.
Currently, I'm not enabling the clock eviction thread. Will enable that later in the day.
NOTE: Don't MIX DDL & DML statements. Delimit them with either a
COMMIT or a ROLLBACK. For the record, SELECT FROM tabtab; is counted as
a DDL statement, because it gets a READ lock on global.tabtab.
The console client can accept input directly from a file as such:
./console_client < SQL_COMMANDS.TXT
ps. plz. do the timing calculation manually this time. I'll be adding
the time taken thing in the next release. Maybe tomorrow. Always for
testing, compile with release=1 vdebug=0 opt=1
If it crashes, then enable debugging, and post the dbug ouput on the group, and the SQL statement which caused it.
--
-Dhruv Matani.
http://www.geocities.com/dhruvbird/"The biggest room is the room for improvement."
-- Navjot Singh Siddhu.