Re: how to hard code @ sign in RPGLE program with out having to worry about CCSID

Joep Beckeringh wrote:

...
- In my test program I did not specify CCSID(*CHAR:*JOBRUN); that would imply that the character data was assumed to be in the mixed-byte CCSID related to the job CCSID.
- When I ran my test program in a job with CCSID 37, the Unicode characters were converted to the equivalent characters in CCSID 37. In a job with CCSID 500 they were converted to 500.

Intuitively I expected conversion to job CCSID, and that is what I found, but it is not what the manual says. According to the manual I should have used CCSID(*CHAR:*JOBRUN) to get the results I got, but I did not specify it.

CCSID 37 is an SBCS (single-byte) CCSID. It has a related DBCS (double-byte) CCSID (I think it's 835) and a related mixed-byte CCSID (I think it's 937). CCSID 937 is a combination of CCSIDs 37 and 835. It uses CCSID 37 for the SBCS data and CCSID 835 for the DBCS data. It uses "shift" characters x'0e' and x'0f' to shift in and out of double byte mode.

If you don't have any x'0e' or x'0f' in your data, then you would have only SBCS data and it would be treated as CCSID 37 data and you wouldn't need CCSID(*CHAR : *JOBRUN).

CCSID(*CHAR : *JOBRUN) is only necessary when the data contains x'0e' or x'0f' that are just ordinary characters. Without *JOBRUN, they are treated as shift characters and x'0e' causes the system to think that there are DBCS characters following the x'0e'.