Thanks for Myrna's answer. But I am asking another function. Currently, I am using Derby in the memory mode.
I need to know the table name list which has data changes during a transaction. So I can export them into csv.
But there is not any information inside derby to tell me this update history. So do I need to manually record the
table names which has really changed during a transaction?
 
Thanks,
Clark

在 2011年3月18日 上午8:49,Myrna van Lunteren <m.v.lunteren@gmail.com>写道:
2011/3/17 鲍少明 <iokerica@gmail.com>:
> I have checked the system tables derby using. I don't find any information
> table to track the modify time of the tables. So does it mean I need to
> write some code to manually track it. And if it's roll-backed, I need to
> clean it up?
>
> 2011/3/14 鲍少明 <iokerica@gmail.com>
>>
>> Hello group,
>> Is there any easy way to track the tables which have been
>> updated/deleted/inserted in one transaction?
>> I know that I can add some code in my db operation code to track the
>> status.
>> But if I need to track some cascade delete or procedure call, I will need
>> to write some ugly code to track them.
>>
>> Is there any easy way to get this done? Any suggestion will be helpful.
>>
>> Thanks,
>> Clark Bao
>

Derby manages all deletes/inserts/updates during a transaction,
including adding of tables...If something fails in the middle, the
entire transaction should get rolled back.
Have you found a situation where this didn't work as you expected?
Perhaps you need to look at your transaction logic...
Perhaps I'm misunderstanding the situation.

Maybe all you want is to compare the contents of SYS.SYSTABLES for
your schema before and after the transaction?

Perhaps you can find more of what you're looking for using the XPLAIN
functionality that was introduced in 10.6; see the Reference Guide
(http://db.apache.org/derby/docs/10.6/ref/ref-single.html) and/or
tuning guide (http://db.apache.org/derby/docs/10.6/tuning/tuning-single.html).

HTH,
Myrna