Re: Manipulating a decoded UTF8 string

Hi, Barbara,
Thanks for clarifying ... :-)
Mark S. Waterbury

On Tuesday, August 18, 2020, 10:06:41 PM EDT, Barbara Morris <bmorris@xxxxxxxxxx> wrote:

On 2020-08-18 4:57 p.m., Mark Waterbury wrote:

...
Note that if you are doing anything "tricky" like declaring a variable to be a certain CCSID on the "declare" statement (or D-spec), you cannot then use a literal in the search, such as coding "%scan('^XZ', base64decoded) " because you do not know what the compiler will decide should be the correct CCSID to use for any "literals" coded in your program.  Normally, I think it defaults to the job's CCSID of the job in which the program was compiled ... :-o
...

Literals are saved in the CCSID that the source file is read in.
Normally, the source is read in the CCSID of the main source file, but
you can use the TGTCCSID parameter to specify a different -EBCDIC-
CCSID, or *JOB.

But by default, RPG interprets the literal in the CCSID of the job the
program is running in.

If you code CCSID(*EXACT) in the H spec, RPG will interpret saved
literals correctly.

Another way to have literals interpreted correctly is to code them as
UCS-2 literals, such as %UCS2('My string').

But %SCAN doesn't allow type or CCSID difference for the operands, so
you'd have to use %CHAR or %UCS2 to get the other operand to the same CCSID.

    %scan(%UCS2('^XZ') : %UCS22(base64decoded))