Prev: how to view a vertical table in horizontal way
Next: REplication of DB2 V9.7 tables with XML columns
From: Serge Rielau on 25 Jan 2010 07:29
On 1/24/2010 8:25 PM, Hardy wrote:
> On 1月25日, 上午6时46分, Serge Rielau<srie...(a)ca.ibm.com> wrote:
>> On 1/24/2010 7:11 AM, Hardy wrote:> On 1月23日, 下午2时50分, Serge Rielau<srie...(a)ca.ibm.com> wrote:
>>>> Indeed. Wait until you see FP2. More good stuff. :-)
>>
>>>> --
>>>> Serge Rielau
>>>> SQL Architect DB2 for LUW
>>>> IBM Toronto Lab
>>
>>> wow~ The curiority can kill a cat. Before you can give me a ppt, a
>>> question, is there any plan for something like Oralce pipeline
>>> feature?
>>
>> Do you mean pipelines table functions?
>> If I understand them correctly they are a pretty good match to a Java or
>> C Table function in DB2 which is also pipelined.
>>
>> Cheers
>> Serge
>>
>> --
>> Serge Rielau
>> SQL Architect DB2 for LUW
>> IBM Toronto Lab
>
>
> yes, by this way database can process query in a more memory-saving
> and even potentialy more efficent way. Serge, could you give some
> example to use java/c table function to pipeline the data? Thanks.
There should be examples in sqllib/samples/c and java

Essentially the table function gets called with a "mode". So the body
typically has a switch statement over the mode.
Modes are: FIRST, FINAL and "regular".
So every invocation the function generates one row.
In newer versions of DB2 that can also be doe in parallel in a DPF
environment.
An Oracle pipelined table function essentially is a loop where rows are
individually returned. So teh difference is that the loop is within the
body while in DB2 the loop is driven by the system. Both operate
"one-row-at-a-time".

Cheers
Serge

PS: There are surprisingly few uses of PL/SQL pipelined functions that I
see which is why they are low priority from an Oracle application
enablement point of view. E.g. defaults for function parameters are much
more pervasive.




--
Serge Rielau
SQL Architect DB2 for LUW
IBM Toronto Lab

From: Hardy Wang on 25 Jan 2010 09:09

"Serge Rielau" <srielau(a)ca.ibm.com> 写入消息
news:7s5h5aFe4nU1(a)mid.individual.net...
> On 1/24/2010 8:25 PM, Hardy wrote:
>> On 1月25日, 上午6时46分, Serge Rielau<srie...(a)ca.ibm.com> wrote:
>>> On 1/24/2010 7:11 AM, Hardy wrote:> On 1月23日, 下午2时50分, Serge
>>> Rielau<srie...(a)ca.ibm.com> wrote:
>>>>> Indeed. Wait until you see FP2. More good stuff. :-)
>>>
>>>>> --
>>>>> Serge Rielau
>>>>> SQL Architect DB2 for LUW
>>>>> IBM Toronto Lab
>>>
>>>> wow~ The curiority can kill a cat. Before you can give me a ppt, a
>>>> question, is there any plan for something like Oralce pipeline
>>>> feature?
>>>
>>> Do you mean pipelines table functions?
>>> If I understand them correctly they are a pretty good match to a Java or
>>> C Table function in DB2 which is also pipelined.
>>>
>>> Cheers
>>> Serge
>>>
>>> --
>>> Serge Rielau
>>> SQL Architect DB2 for LUW
>>> IBM Toronto Lab
>>
>>
>> yes, by this way database can process query in a more memory-saving
>> and even potentialy more efficent way. Serge, could you give some
>> example to use java/c table function to pipeline the data? Thanks.
> There should be examples in sqllib/samples/c and java
>
> Essentially the table function gets called with a "mode". So the body
> typically has a switch statement over the mode.
> Modes are: FIRST, FINAL and "regular".
> So every invocation the function generates one row.
> In newer versions of DB2 that can also be doe in parallel in a DPF
> environment.
> An Oracle pipelined table function essentially is a loop where rows are
> individually returned. So teh difference is that the loop is within the
> body while in DB2 the loop is driven by the system. Both operate
> "one-row-at-a-time".
>
> Cheers
> Serge
>
> PS: There are surprisingly few uses of PL/SQL pipelined functions that I
> see which is why they are low priority from an Oracle application
> enablement point of view. E.g. defaults for function parameters are much
> more pervasive.
>
>
>
>
> --
> Serge Rielau
> SQL Architect DB2 for LUW
> IBM Toronto Lab
>

um, I'll study the samples. my experience focused much more on DBA things
than on developer things. I should do more homework:)

First  |  Prev  | 
Pages: 1 2 3
Prev: how to view a vertical table in horizontal way
Next: REplication of DB2 V9.7 tables with XML columns