Is it necessary to commit after insert statement ?

[former_member220286]

Hi Gurus ,

Hi I have a requirement that I need to update by ztable with insert statement in an fm .I just want to know that is it necessary that after insert we need to write the commit statement .Please suggest .

DATA(lv_length) = strlen( lv_sernr ).
IF lv_length > 10 .
wa_meterno-meter_no = lv_sernr.
SELECT SINGLE meter_no FROM zmeter_no INTO @lv_meter_no WHERE meter_no = @wa_meterno-meter_no.
IF sy-subrc <> 0 .
INSERT zmeter_no FROM wa_meterno.
CLEAR wa_meterno.
ENDIF.

Regards

Deep

[FredericGirod]

If you do a commit, you will saved all the database modification done in your program. (so maybe other done before)

If you do nothing, depending of the type of FM, the commit will be implicite.

That means, you commit only if you really need to have the data stored in the database, because you will use it just after this INSERT statement.

[joltdx]

Hey, frdric.girod, make that an answer instead of a comment, so I can upvote it... 🙂

[Sandra_Rossi]

After a database update, as per database principle, you must COMMIT or ROLLBACK. Doing nothing is non-sense, or it means that you know you are part of a whole process which ends with a COMMIT or ROLLBACK. Search the Web for more information (ACID for instance, atomicity, consistency, isolation, durability)

[joltdx]

sandra.rossi Yes, I think you and I agree here. In this case i think the question is about a function module (if that is what "insert statement in an fm" is), and then it should most likely be a "part of a whole process". I prefer FM's, and methods, to not COMMIT everything they do. It should be done but in a controlled manner, which also would give the possibility of a ROLLBACK if needed.

Or did I misunderstand you?

[Sandra_Rossi]

Jörgen Lindqvist Well I don't understand what you mean. I don't want to talk about FMs and methods, because I don't know the context of the OP question (I let you the responsibility of "most likely"). The only thing I can say is that there must be a COMMIT and a ROLLBACK somewhere. Maybe that answers the question, maybe not, I don't know.

[joltdx]

Yes, sandra.rossi, I see, and we agree. 🙂

These questions are not always clear to understand, but OP says "update by ztable with insert statement in an fm". In my interpretation it's a question about a function module, that's why I was asking... But you never know... 🙂

[Abinathsiva]

Hi,

usually for inserting data to ztable - Insert statement is sufficient. modify also can be used, commit is not required.

[Sijin_Chandran]

Hi,

Must say the information you have mentioned here is not sufficient for this, but let me try answering this by trying to cover all the aspects.

You have just mentioned INSERT in an FM, now here are the points :

1. If the FM is a SMOD EXIT FM you don't need any COMMIT, because Standard Transactions COMMIT will take care of everything.

2. You should never use COMMIT inside an EXIT/BADI/Enhancements etc because it will affect the Data consistency. In case you need to update DATA in a Custom Table , create a FM with UPDATE Task and put the INSERT statement inside that FM and call this FM in the Enhancement. This is similar to the Point 1 only.

3. Apart from this if you are doing an INSERT statement in a Custom Program then you can call COMMIT after INSERT.

Hope this answers.

Thanks,

Sijin

[Sijin_Chandran]

2. You should never user COMMIT inside an EXIT/BADI/Enhancements etc because it will affect the Data consistency. 

I feel like I should explain this as well. So here we go.

Suppose in an EXIT for PO you have wrote a code for updating a ZTABLE with respect to a PO. Then that Table should be updated only if we are sure that PO itself gets updated to DB successfully.

This can be made sure or taken care by updating the ZTABLE in the Update Task FM. On the contrary if you use COMMIT inside EXIT for Table INSERT, it will independently update that ZTABLE without waiting for the PO to COMMIT to DB. Resulting to an inconsistent Data.

Hope I was able to make you understand.