Re: bzr commit into mysql-next-mr-bugfixing branch (jorgen.loland:3242) WL#4800

[Jorgen Loland]

Guilhem,

I think patching a broken diff-file has played a trick on you regarding the 
questions below (the comments I have removed are valid). I will give commit 
another try so you can get a proper bundle.

On 01/04/2011 03:03 PM, Guilhem Bichot wrote:
> Hello,
>
> First, I'm grateful that you took the time to fix all issues; I see
> there were quite a few in my own code :-/
...
>> === modified file 'mysql-test/r/optimizer_trace_no_prot.result'
>> @@ -1561,12 +1688,12 @@ explain extended select * from t1 where },
>> {
>> "attaching_conditions_to_tables": {
>> - "original_condition": null,
>> + "original_condition": "<nop>((`test`.`t1`.`s1` > (/* select#2 */
>> select min(`test`.`t1`.`s2`) from `test`.`t1`)))",
>
> It's not obvious in this diff, but this block (like 1641 in the result
> file after the patch) belongs to select#2 (subquery) so it's not normal
> that its "original_condition" is now "<nop>(t1.s1>(select#2 etc": that
> is surely a condition for select#1. There is a problem.
>
>> "attached_conditions": [
>> {
>> "database": "test",
>> "table": "t1",
>> - "attached": null
>> + "attached": "<nop>((`test`.`t1`.`s1` > (/* select#2 */ select
>> min(`test`.`t1`.`s2`) from `test`.`t1`)))"
>> }
>> ] /* attached_conditions */
>> } /* attaching_conditions_to_tables */
>
>> @@ -3436,12 +3647,12 @@ concat(c1,'y') IN
>> },
>> {
>> "attaching_conditions_to_tables": {
>> - "original_condition": null,
>> + "original_condition":
>> "(<in_optimizer>(concat(`test`.`t1`.`c1`,'x'),concat(`test`.`t1`.`c1`,'x')
>> in (/* select#2 */ select left(`test`.`t2`.`c2`,8) from `test`.`t2`))
>> and
>> <in_optimizer>(concat(`test`.`t1`.`c1`,'y'),concat(`test`.`t1`.`c1`,'y')
>> in (/* select#3 */ select left(`test`.`t2`.`c2`,9) from `test`.`t2`)))",
>> "attached_conditions": [
>> {
>> "database": "test",
>> - "table": "t2",
>> - "attached": null
>> + "table": "t1",
>
> This change from t2 to t1 is not normal: this block (line 3651 of the
> result file after the patch) is for select#2 (subquery) which doesn't
> contain t1.
>
>> + "attached":
>> "(<in_optimizer>(concat(`test`.`t1`.`c1`,'x'),concat(`test`.`t1`.`c1`,'x')
>> in (/* select#2 */ select left(`test`.`t2`.`c2`,8) from `test`.`t2`))
>> and
>> <in_optimizer>(concat(`test`.`t1`.`c1`,'y'),concat(`test`.`t1`.`c1`,'y')
>> in (/* select#3 */ select left(`test`.`t2`.`c2`,9) from `test`.`t2`)))"
>> }
>> ] /* attached_conditions */
>> } /* attaching_conditions_to_tables */
>
>> @@ -4150,13 +4473,117 @@ select * from t1 where (t1.a,t1.b) not i
>> "refine_plan": [
>> {
>> "database": "test",
>> - "table": "t2",
>> + "table": "t1",
>
> not normal
>

-- 
Jørgen Løland | Senior Software Engineer | [phone]
Oracle MySQL
Trondheim, Norway