Re: Where is a temp tables DDL stored?

"Mark C. Stock" <mcstockX_at_Xenquery .com> wrote in message news:Ze6dncQGd-NUmrLeRVn-iw_at_comcast.com...
>
> "Ed Prochak" <ed.prochak_at_magicinterface.com> wrote in message
> news:1127152935.283980.15370_at_g14g2000cwa.googlegroups.com...
>>
>> mikharakiri_nosp..._at_yahoo.com wrote:
>>> Jonathan Lewis wrote:
>>> > Just playing devil's advocate, but why couldn't
>>> > the database reverse engineer the corrected
>>> > DDL from the data dictionary as the dictionary
>>> > is updated ?
>>>
>>> To extend this idea further, I always wondered why the concept of DDL
>>> is needed at all? Consider
>>>
>>> -- create 100 tables TEST1...TEST100
>>> insert into user_tables -- or, more concervatively, tab$
>>> as select 'TEST'||rownum table_name, ...
>>> from dual connect by rownum < 100
>>>
>>> insert into user_tab_columns -- or, more concervatively, col$
>>> ...
>>>
>>> commit;
>>
>> I'd say its the side-effect issue. With the above approach, you insert
>> into one table and Voila a new table exists.
>>
>> But also note: using that method you can "create" a table with NO
>> columns. And what about the other side effects of CREATE TABLE:
>> allocating space, granting permissions, etc.?
>>
>> So direct inserts in the Data dictionary table are not a great idea,
>> even if that is what the DBMS does in the background.
>>
>> ed
>>
>
> just consider DDL to be the data dictionary API -- just parsed instead of
> parameterized
>
> ++ mcs
>

make that a '(semi-)standardized data dictionary API', since all RDBMS data dictionaries are not structured the same

++ mcs Received on Mon Sep 19 2005 - 13:48:47 CDT