Strange Oracle error 5230

[jvv]

I have a very very diffuclt problem. If I try to create a record I got error message 5320. But I am the only user, and I do not lock optimistic.
It does not not matter if the table is empty or not, the error will alway occur.

The code that causes the error is:

....
on go of frame:
assign < fields >
release < table >
return rowid(< table >)
end.

....
in the main block:
create < table >

So the whole procedure is one huge transaction. I have debugged the code and I discoverd that the release statement fires the write trigger (as expected). The write trigger is run properly, but then, returning to the release statement the error 5320 pops up.

To make it more difficult: I have the same procedure for other tables, and these work fine!



[Edited by progresstalk on 30 Nov 2000 at 10:56 AM]

 

[rf08]

Stranger Oracle error 2049

While trying to create an Oracle table row and purposely not setting the value on a column with a not null requirement the transaction will fail and DOES NOT do a complete UNDO (Progress). A rerun of the program to post the same record results in the 2049 message thus the row create failure is limited to one attempt only because the composite values of ALL unique keys in the table were the same. Creating other row with different key values is not restricted. This (uncreated record lock out) problem does not go away until the oracle dataserver in bounced (shutdown and started).

We are using explicit RELEASE statements with NO-ERROR key word on CREATE, ASSIGN and RELEASE in conjuction with the system logical variable error-status:error. This combination suppresses most Oracle failure messages except the above.

Our configuration is somewhat unique

Progress 8.1A03 on DEC Alpha OPENVMS
with Oracle 7.3 on the same machine which is then DBlinked to an Oracle 8 DB on a SUN 450 (2.6). Progress 8 cannot access directly an Oracle 8 DB therefore we created a shell Oracle 7 DB with views to the 8 DB.

Roger