Unicode/OO issues with HR_INFOTYPE_OPERATION in separate LUW

Question

Hi,
I've created a separate program that uses FM HR_INFOTYPE_OPERATION in order to allow infotype changes while still being in the PAI of a PA30 infotype operation. 
Because I have so many requirements for infotype changes to be triggered from PA30 like this, I decided to make the program dynamic as opposed to defining individual versions per infotype required.
I did this by basically putting the FM parameters as the selection screen entries, defining RECORD and SECONDARY_RECORD as strings NO-DISPLAY and then casting the PNNNN structure onto the string contents in the START-OF-SELECTION, like so:
* Dynamic data object of unknown line type
  CONCATENATE 'P' p_infty INTO lv_pnnnn.
  CREATE DATA gdo_data1 TYPE (lv_pnnnn).
  ASSIGN gdo_data1->* TO <fs_record>.
  <fs_record> = p_record.
This has worked for me until very recently for some reason.  Now when I submit the program, I get the error "ABAP/4 processor: UC_OBJECTS_NOT_CONVERTIBLE" in the job log at the point when I make the statement:
<fs_record> = p_record.
Where p_record is the string.
In an unicode/OO environment, any idea on how I can submit a PNNNN structured record through the selection screen of a program  that will dynamically reapply it's structure before submitting to HR_INFOTYPE_OPERATION?
More pointedly, it looks like I'm successfully transferring from PNNNN structure to string, but dumping when I try to transfer from string back to PNNNN structure.  How can I deal with char and num fields when transferring back to the structure without having to do it field by field?
Many thanks
null

Answer 1

I need it to be enabled in this system so it's checked.  Thanks for the suggestion though.