Re: Future of RPG was: Open Access for RPG

Aaron

In a PF, each field can have a CCSID. In message descriptions, this is an option, too.

The idea is not to have any character literals in your source code - things like the strings you might use to convert from lower- to upper-case. Even things like delimiters - a tilde, for example, will get mangled. Have them all either in message descriptions or read them in from a file. I don't think data areas can be used for this, there's no CCSID on a CRTDTAARA command. I could be wrong, of course.

If you flag a message description as CCSID 37, then when it is read into your program, it is automagically converted to the job CCSID using internal conversion stuff. Probably what iconv uses. At least that's what I'm told.

It doesn't matter what CCSID your source code is in this scenario. Because there are no CCSID-specific items in the code.

Cool, eh? If it works!!!

Vern

On 7/29/2010 8:26 AM, Aaron Bartell wrote:

Vern, what do you mean "flagged with a CCSID"? Where is this flag?

Aaron Bartell
http://mowyourlawn.com
http://mowyourlawn.com/blog/



On Thu, Jul 29, 2010 at 8:08 AM, Vern Hamberg<vhamberg@xxxxxxxxxxx> wrote:

Aaron

It should not be necessary to change CCSID of source and recompile -
that is one solution, but it is needed only if you use character
literals in the source. We have done that here with one of our products.

As I understand it - and Bruce Vining can say more and correct me - if
you have text in PFs and MSGFs and it's flagged with a CCSID, the system
just converts it as needed. This is very cool and the secret to simple
national language processing.

That's a very simplistic description, but I think it's essentially
correct. But please, others need to weigh in here.

Vern