Hi!
>>>>> "Yuri" == Yuri Dario <mc6530@stripped> writes:
Yuri> Hi,
Yuri> now OS/2 code is failing only 6 tests of the 104 available (skipping
Yuri> replication tests) :-))
Yuri> Here is the latest report:
Yuri> 1) alter_table.test: check table reports
Yuri> test.t1 optimize status Table is already
Yuri> up to date
Yuri> instead of
Yuri> test.t1 optimize status OK
Very strange; The message is ok, but I don't understand why this is
different from the standard one...
Yuri> 2) func_test: at line 16
Yuri> "Det hõr õr svenska" regexp "h[[:alpha:]]+r" "aba"
Yuri> regexp "^(a|b)*$"
Yuri> 0 1
Yuri> instead of
Yuri> "Det hõr õr svenska" regexp "h[[:alpha:]]+r" "aba"
Yuri> regexp "^(a|b)*$"
Yuri> 1 1
Are you using our regexp library or a one that is in OS/2 ?
Are you using latin1 as the character set ?
Yuri> 3) merge: 'select * from t3 order by a desc' is failing, reporting
Yuri> a b
Yuri> 1 Testing
Yuri> 1 Testing
Yuri> 1 Testing
Yuri> 1 Testing
Yuri> 1 Testing
Yuri> 1 Testing
No comments; I would need a trace to give any hints about this.
Yuri> 4) select.test: at line 1458 (probably some rounding)
Yuri> 37 12543 309394878010 -0.0000 464091
Yuri> instead of
Yuri> 37 12543 309394878010 0.0000 464091
I agree. I have now added 'abs' around the result to fix this.
Yuri> 5) show_check.test: same as alter_table, plus
Yuri> 59,61c59,61
Yuri> < t1 0 PRIMARY 1 f1 A NULL NULL
Yuri> NULL
Yuri> < t1 0 PRIMARY 2 f2 A NULL NULL
Yuri> NULL
Yuri> < t1 0 PRIMARY 3 f3 A NULL NULL
Yuri> NULL
Yuri> ---
>> t1 0 PRIMARY 1 f1 A 1 NULL
Yuri> NULL
>> t1 0 PRIMARY 2 f2 A 3 NULL
Yuri> NULL
>> t1 0 PRIMARY 3 f3 A 9 NULL
Yuri> NULL
This also looks ok.
This is basicly the same problem as the alter test;
It looks like you have run the CHECK two times on the tables.
Did you run the test 'standalone' or with 'mysql-test-run --force' ?
Yuri> 6) status.test: different locking count
Yuri> 5,6c5,6
Yuri> < Table_locks_immediate 4
Yuri> < Table_locks_waited 0
Yuri> ---
>> Table_locks_immediate 3
>> Table_locks_waited 1
This is strange; If you are using just one client, you should never
get table_locks_waited.
Yuri> Actually, I think merge error is on top in the bug list. From the
Yuri> mysqld.trace, I don't understand where to look, at least without
Yuri> debugging the do_select().
The thing to do is to do a trace on OS/2 and a trace on Linux and
compare those.
Can you ftp a trace where you ONLY did run the merge test to
ftp://support.mysql.com/pub/mysql/secret
so that I can take a quick look at this ?
Regards,
Monty
