On Tue, 21 Aug 2018 17:30:36 +0000, Seymour J Metz wrote: >Be careful what you ask for; you might get it. Unless you tell the access >method how to translate, it likely to trip up the application. Even if you >have both to and from CCSID on the DD statement, > According to the JCL Ref., one goes on the EXEC (what the program processes), the other on the DD (what's on the external medium). I suspect MBCS is unsupported, and it mentions a default of 7-bit ASCII!
There are UNIX syscalls to control translation. "od" and "cksum" (e.g.) rely on them. IBM fixed my APAR so they work (FSVO) right. >... there are potential pitfalls. I'd much rather leave the application in >control. Something like DCBE and processing options to specify > > Translate/raw > Application CCID > Query file CCSID > Others I've overlooked > "65535" (Yaay! not "0") in JCL means "raw". I'd expect JCL merge dominance rules to apply. Does filetagging the UNIX Program Object have effect similar to CCSID on the EXEC statement? Only 1047 and 819 are supported by z/OS UNIX. Spool files ought to be taggable. CHARS is some sort of option orthogonal to CCSID. -- gil ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
