How to capture PKID after insert in Oracle

Question

All--
Please help.
This is a general question about Oracle in general, in the hope that someone has solved this problem.
I want to capture the PKID after an insert into Oracle.
I expect that this is just a syntax issue.
I use DataAdapter.Update() passing in a DataTable.
It looks like...
...the trigger is using the sequence correctly...
...the PKID value is provided for insert correctly...
...the insert occurs correctly...
...the output parameter is set by the SP...
...and then the DataAdapter trys to fill the underlying DataTable's PKID column and thats where the problem occurs...
...so, I think,...
...either the cast from OracleType Number (from the SP) to DotNetType decimal (from the DataTable) is failing...
...or...
...for some unknown reason Oracle is actually passing <10007>, with the brackets included, after the insert and casting that to decimal fails... (but, this seems unlikely)...
...or...
...something else?
Any ideas?
(See below for details.)
...this is the error...
Code:
System.ArgumentException: System.InvalidCastException: Specified cast is not valid. at System.Convert.ToDecimal(Object value) at System.Data.Common.DecimalStorage.Set(Int32 record, Object value) at System.Data.DataColumn.set_Item(Int32 record, Object value)Couldn't store <10007> in PKID Column. Expected type is Decimal. at System.Data.Common.DbDataAdapter.Update(DataRow[] dataRows, DataTableMapping tableMapping) at Oracle.DataAccess.Client.OracleDataAdapter.Update(DataRow[] dataRows, DataTableMapping tableMapping) at System.Data.Common.DbDataAdapter.Update(DataTable dataTable) at DLSLOD.Private.Business.Entity._LodAccesscode.Save(OracleConnection p_OpenConnection) in c:\vsswap3\wadprojects\dls\dlslod_01\dlslod\private\business\entity\_lodaccesscode.cs:line 715 at DLSLODAdmin.TestLodAccesscode01.SaveNewRow_Click(Object sender, EventArgs e) in c:\vsswap3\wadprojects\dls\dlslod_01\dlslodadmin\testlodaccesscode01.aspx.cs:line 144
...and the SP looks like this...
Code:
CREATE OR REPLACE PROCEDURE LOD_ACCESSCODE_INS
P_PKID OUT LOD_ACCESSCODE.PKID%TYPE,
P_DEPT_PKID IN LOD_ACCESSCODE.DEPT_PKID%TYPE,
P_NAME IN LOD_ACCESSCODE.NAME%TYPE,
P_NOTE IN LOD_ACCESSCODE.NOTE%TYPE,
P_CHANGEACCT IN LOD_ACCESSCODE.CHANGEACCT%TYPE,
P_CHANGEDATE OUT LOD_ACCESSCODE.CHANGEDATE%TYPE
IS
BEGIN
INSERT INTO LOD_ACCESSCODE
DEPT_PKID,
NAME,
NOTE,
CHANGEACCT
VALUES
P_DEPT_PKID,
P_NAME,
P_NOTE,
P_CHANGEACCT
BEGIN
SELECT LOD_ACCESSCODE_SEQ.CURRVAL INTO P_PKID FROM DUAL;
END;
BEGIN
SELECT CHANGEDATE INTO P_CHANGEDATE FROM LOD_ACCESSCODE WHERE PKID = P_PKID;
END;
END LOD_ACCESSCODE_INS;
...and the trigger that fills the PKID is this...
Code:
TRIGGER LOD_ACCESSCODE_TR_SEQ
BEFORE INSERT
ON LOD_ACCESSCODE
FOR EACH ROW
BEGIN
SELECT LOD_ACCESSCODE_SEQ.NEXTVAL
INTO :NEW.PKID
FROM DUAL;
END;
...and my sequence looks like this...
Code:
CREATE SEQUENCE LOD_ACCESSCODE_SEQ INCREMENT BY 1 START WITH 10001 MINVALUE 10001 MAXVALUE 9223372036854775807

Answer 1

All--
FYI, this is a follow-up to my post above.
After much Googling, I found no helpful answer to this issue.
Therefore, I have implemented a workaround.
Here is the situation and current solution.
--I do a "select * from SomeTable where 1 = 2" to get the schema from the database into an empty DataTable.
--Internally, DotNet converts from Oracle datatypes to DotNet datatypes. Therefore, the DataTable is created with typed columns. This is great so far.
--Note that in the case of a column with Oracle datatype "NUMBER" DotNet creates a column in the DataTable with the datatype as "decimal". Fine, this should work. DotNet is now sending the message that an Oracle datatype "NUMBER" can and will be converted automatically to a DotNet datatype "decimal".
--Now, I need to make an insert and then put the resulting OUT parameter from an Oracle column with datatype "NUMBER" into that DataTable column with DotNet datatype "decimal". One would think that this should work because DotNet already allowed this conversion once.
--Not so.
--When Oracle returns a column of datatype "NUMBER" as an OUT parameter from a proc, it cannot be cast internally by DotNet from NUMBER to decimal. It fails with a conversion error, as shown in detail above.
--Now, this would make some sense if it were consistent. For example, it may be that C# disallows narrowing casts in this scenario. However, what is not consistent and does not make sense is that DotNet does allow narrowing conversions when selecting the schema and building the initial DataTable. Also, given that the magnitude of an Oracle NUMBER exceeds that of a DotNet decimal (and no other DotNet numeric datatype will hold all possible values of a NUMBER), then DotNet either must allow narrowing conversions in this case OR we would never be able to map the datatypes from one platform to the other, which would be a serious problem.
--Therefore, since one cannot change the way Oracle is behaving, an adjustment needs to be made in the middle-tier. The datatype of the column needs to be set to some friendly datatype that both DotNet and Oracle can agree upon.
What I now do is grab the DataTable, get the column, and forcibly set the datatype to "string". It seems to work just fine, especially since that DataTable is for simple UI binding so, the datatype of the column is not important.
That's the "solution" for now.
An overview of the code is below.
Please send along any thoughts you may have.
Thank you.
--Mark
...here is what the SP now looks like...
Code:
CREATE OR REPLACE PROCEDURE LOD_ACCESSCODE_INS
P_PKID OUT LOD_ACCESSCODE.PKID%TYPE,
P_DEPT_PKID IN LOD_ACCESSCODE.DEPT_PKID%TYPE,
P_NAME IN LOD_ACCESSCODE.NAME%TYPE,
P_NOTE IN LOD_ACCESSCODE.NOTE%TYPE,
P_CHANGEACCT IN LOD_ACCESSCODE.CHANGEACCT%TYPE,
P_CHANGEDATE IN LOD_ACCESSCODE.CHANGEDATE%TYPE
IS
BEGIN
INSERT INTO LOD_ACCESSCODE
DEPT_PKID,
NAME,
NOTE,
CHANGEACCT,
CHANGEDATE
VALUES
P_DEPT_PKID,
P_NAME,
P_NOTE,
P_CHANGEACCT,
P_CHANGEDATE
BEGIN
SELECT LOD_ACCESSCODE_SEQ.CURRVAL INTO P_PKID FROM DUAL;
END;
END LOD_ACCESSCODE_INS;
...and here is the adjustment that I made in the middle-tier....
Code:
System.Type myType;
myType = System.Type.GetType("System.String");
_CurrentDataTable.Columns["PKID"].DataType = myType;