On Sun, 9 Oct 2022 21:07:57 +0000, Farley, Peter x23353 wrote: >>> >>If it doesn't work it deserves an SR. Can you Edit/View files so tagged with >>ISPF? > >IIRC I have to use the 3.17 Unix directory browser and a "/" to execute "View >ASCII/UTF8" or "Edit ASCII/UTF8", as I am unfamiliar with the 3.17 line >commands to do that (VA or EA maybe? I really need to go look those up and >start using them), but yes, I CAN view or edit them when so tagged. > In my experience, a file tagged 819 or 1208 and FILEDATA is recognized and processid by ISPF; I needn't specify EA or VA.
>>>I originally did not think of using LRECL/RECFM overrides on the SYSEXEC >>>allocation because I thought they would be incompatible with the PATH >>>keywords. >>> >>I'd say you underestimate IBM, but I've had too much unpleasant experience >>overestimating IBM in such matters. BTW, DCB=(LRECL,...) is incompatible >>with PATH. WHY!? > >Good question, but probably not one to which we can get a straight answer. At >a guess, old and crotchety JCL interpretation code that no one wants to touch >(if it ain't broke, don't fix it), while the "outside a DCB" keywords are >probably in newer OCO code that is "easier" (FSVO "easier") to maintain. > I suspect the JCL Converter has no ability to detect a conflict between one option and a suboption of another option. >>> IBM could provide better and more complete examples of accessing Unix >>> files from a TSO or batch task >+1 ... >N t just HLASM SYSLIB, but also Binder SYSLIB and any/all HLL compiler >SYSLIB's, STEPLIB/JOBLIB, et alia: Basically, anywhere a Unix directory can >validly be used as a library. Such usage probably warrants a sub-chapter of >its own, or at least a page or two somewhere prominent. > I believe Binder is exceptional. It (necessarily) supported UNIX files in an intervel when Allocation supported PATH but access methods didn't. In consequence: o Binder does not support non-trivial concatenations containing UNIX PATHs. o Binder ignores FILEDATA and - treats SYSLIN as BINARY - treats SYSPRINT as TEXT. I consider it indolent design that Binder does not issue a Warning if the programmer codes a conflicting FILEDATA. I've whined about that here and an IBM representative (Peter?) has said that if I supply invalid input I should not expect any specific behavior such as a message. That's below the product quality I expect of IBM. o Binder does not use BPAM, I believe in part because BLDL can't deal with UNIX filenames. >Suggested (sub)chapter title: "Using Unix Directories as libraries". JCL >Programmer's Guide perhaps, with sufficient examples to cover all the bases >for both TSO and batch execution. > >Actually, I don’t think I've even looked at the JCL Programmer's Guide in too >many years, so maybe I should go see what's there these days. > >>> ... that show which DD keywords are compatible with Unix allocations, but >>> I won't hold my breath waiting for such to be created. There may be some >>> table(s) somewhere (maybe in the JCL reference manual?) that show >>> compatibility, but it's my fault that I haven't looked for them yet. >>> >>There used to be such a matrix, in the JCL Ref., IIRC. Perhaps it outgrew >>page size and IBM >>simply dropped it in favor of scattered "Relationship to other parameters" >>sentences. -- gil ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
