On Wed, 6 May 2020 20:40:20 +0800, David Crayford wrote: >We've had this conversation many times before. Utilities that use >fopen() support DD:xxx and are not likely to change. > >Gil is being pedantic. But maybe IBM should update the doc? > Pedantic, yes, but I'd be loath to use in code distributed to customers constructs not documented as supported.
I submitted such an RCF years ago saying that many of my peers misuse DD:xxx in this fashion and suggesting a clarification that the Appendix should state clearly that it is not supported. IBM declined, saying (roughly paraphrased), "We document what we support; we don't document what we don't support." Of course; the latter is an enormous set, impractical to enumerate. And I admired the gyrations of a co-worker to avoid using FALSE or TRUE in JCL IF-THEN-ELSE as boolean primaries. They happen to work but are not documented. She was being fastidious. >On 2020-05-06 8:31 PM, Kirk Wolf wrote: >> cat DD: isn't documented as supported? :-) >> >> what about "cp //DD:xxx /dev/fd1" - is that the shell? >> I don't believe shell is closely involved. I'd expect similar behavior if it were invoked from a program by exec(). "DD:xxx" would be elaborated by the C RTL provided that cp invokes fopen(), which is not guaranteed, and /dev/fd/1 is a kernel facility. -- gil ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN