Just found the following note on matalink. Looks like may not be possible

419797.1

Which states

"For Error 2: PAY HR_PERSON_DUPLICATE : A person of this name already exists. Continue?

Both an employee and contact record have the same last name and first name, which causes a Caution in the Professional User Interface (PUI) and causes an error in Web ADI.during upload.


Fact: Caution and warring messages that occur in the PUI are not supported in Web ADI."

This all sounds very odd to me... The api will do the create or update, but in doing so might or might not set a warning flag. The returned status is still 'success', so no exception activity will occur. When you run the code through a sql harness, it all works fine. In fact, unless you specifically coded for the warning situation, you'd not know the warning occurred. So why does Web-ADI take it upon itself to act upon this warning?

One other question: what kind of integrator are you running this with?

As an aside:

Don't take this as a full-blown peer review or anything like that, but from looking at the 'update assignment' wrapper, I see that you are defaulting the input parameters to null...

You might want to consider the general strategy of how to interpret a null input within your wrapper, i.e. is it really a null, or is it a 'do not change' instruction? If you leave your code as is, it will have the effect of trying to set everything to null that you aren't explicitly setting to something.

Where I have had occasion to do this in the past, I built a procedure to convert a null to a 'do not change' value, and the text '[NULL]' to a true null (just in case the user genuinely wishes to set something to null)

I'll change to hr_api.g_varchar2 defaults. Thanks for tip, always appreciated. As all self taught I get into bad habit's until they cause a fail, then bang my head against a wall for half a day or so wondering what went wrong. So hopefully that tip will save me headache.

It's a create integrator

No worries, always happy to try and help someone who is trying for him/herself to solve a problem.
I do think however that the changing of the default to g_varchar2 might not be enough. If the integrator passes a null from the spreadsheet input, then it will override the default anyway - so it may be that you need to explicitly trap nulls and convert them to g_varchar2 as appropriate.


OK, if possible can you post some more details of the setup of the integrator, e.g. Screenshots or FNDLOAD type extracts?

Been using the HR Seeded integrator template to set up the following:

CT - You were right just using the default to g_varchar2 didn't prevent fields being updated to null, I had to catch and convert.

Cheers again