On Feb 2, 2022, at 07:45:40, ITschak Mugzach wrote: > > Co-posted to IBM-MAIN and TSO-REXX I write a small rexx program and place > it in a directory. it onl;y have the following : > /* rexx */ > Say 'Hello World' > > I start the exec with ./myexec.rex while pwd shows the correct directory > and ls -l shows the exe. it returns BPXW0000I Exec not found > > What am I missing? > Appending an old discussion from MVS-OE. My perception is that William Schoen's prosal would not fix the problem, but break it further. I have seen even worse behavior when exec() correctly specifies a path to a Rexx exec which is not in PATH, but PATH contains a file which is not Rexx.
A proper fix would make the "path" argument work correctly for Rexx > On Jun 26, 2018, at 17:36:18, William Schoen <w...@us.ibm.com> wrote: > > You remember correctly. > I forgot about that behavior. > It was not intended and should have been fixed some time ago. > If you find it still does that please open a PMR. > As I recall this should have been fixed such that the first argument and > program name must resolve to the same file node. > > Bill Schoen > > MVS OpenEdition <mvs...@vm.marist.edu> wrote on 06/26/2018 05:47:12 PM: > >> From: Paul Gilmartin <paulgboul...@aim.com> >> To: mvs...@vm.marist.edu >> Date: 06/26/2018 05:48 PM >> Subject: Re: Fwd: REXX shell script - start a flame war? >> Sent by: MVS OpenEdition <mvs...@vm.marist.edu> >> >> On 2018-06-26, at 14:54:08, William Schoen wrote: >> >>> I misunderstood. It still works that way. pathname and arg_stem.1 > must >>> match. >>> >> That was not my experience. If arg_stem.1 resolved to a valid Rexx >> EXEC it executed that instead. Sometimes with astonishing results. >> >>> Whether it is rexx or C, the first argument must match the pathname. >>> >> Again, not my understanding. C always executes the pathname program, >> passing it the caller's argv[0]. >> >>> Conventionally, arg[0] contains the program name. REXX requires that. > I >>> won't go into why. >>> >> Many years ago, you said there simply weren't enough arguments in the >> interface to the interpreter. That could have been fixed (alternate >> entry point?) I suspect the TSO folks were offended by the requirement >> to support for OMVS and dug in their heels wherever they could. >> Likewise you said in that era that you had campaigned for TRACE and >> message output to go to stderr (as Regina does by default) but your >> equest was denied. >> >> The Rexx base defended their territory and the customers suffered.' -- gil ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN