On 9/1/2012 6:38 AM, Vernon Hamberg wrote:
Thanks for clarifying - now is this true for source you create on a
US English machine and open in a British system? Or using pipes in CL
for *CAT, etc.? I believe it is best to use invariant characters when
distributing source - again, I welcome confirmation or the
As long as source files are tagged with the correct CCSID, they will be trans-coded correctly by the database engine when they are copied (I think you might need to use FMTOPT(*MAP), but I'm not sure).
Once they are compiled, invariant characters used as op-codes don't seem to pose a problem. We have plenty of CL programs that use pipe's instead of *CAT that operate in many different locales (UK, Japan, Germany, Italy, etc) without a problem.
I'm not sure what the impact of using invariant characters in column names in SQL tables would be though ... since a lot of SQL stuff seems to be done on the fly. I'm 99% sure it would screw up dynamic SQL (SQL statements that are built on the fly).