On Fri, 26 Jun 2026 05:58:11 +0000, Farley, Peter wrote: >I would think that a file opened with O_APPEND (same as for DISP=(MOD,…) in >JCL) should be allocated exclusively to the first process where the open >succeeds and that process should keep exclusive control of that file/stream >until it issues a close. All other processes that issue an open for the same >file/stream after the first one succeeds should get an error until the first >process releases control via close. > Cite? More definitive than "would think".
No, Not "O_APPEND (same as for DISP=(MOD,…) in JCL)." UNIX is not MVS. It allows filenames longer than 44 characters and program objects residing in the sane directory as REXX execs. Can you imagine how it can do that!? It's quite ordinary for a UNIX background job to maintain a log with O_APPEND and impatient programmers to monitor that with "tail -f" <https://www.ibm.com/docs/en/zos/3.2.0?topic=descriptions-tail-display-last-part-file> Would that work with DSFS? Where's the DSFS User's Guide where I might seek the answers to suchh questions? >From: Paul Gilmartin >Sent: Thursday, June 25, 2026 8:42 PM > ><https://urldefense.com/v3/__https://www.ibm.com/docs/en/zos/3.2.0?topic=functions-rexx-io__;!!Ebr-cpPeAnfNniQ8HSAI-g_K5b7VKg!N6mSOxzwWkYazRF1K-8ir8aX0kvobaWND8Ka4KfzkVqqdbsJXoQQXB2JDzLV-K3UHVk0WP_G8hRaZZEwtjcpyBLiEmaQ5-pf6XXwYV_e$<https://urldefense.com/v3/__https:/www.ibm.com/docs/en/zos/3.2.0?topic=functions-rexx-io__;!!Ebr-cpPeAnfNniQ8HSAI-g_K5b7VKg!N6mSOxzwWkYazRF1K-8ir8aX0kvobaWND8Ka4KfzkVqqdbsJXoQQXB2JDzLV-K3UHVk0WP_G8hRaZZEwtjcpyBLiEmaQ5-pf6XXwYV_e$>> > [I didn't say that. Is it tracking us?] > In REXX, the concept of persistent and non-persistent > > (or transient) streams is closely tied to how these input and > > output functions interact with the stream. If a stream is > > explicitly open d using the STREAM() function, it becomes > > persistent, meaning that it retains its position and state > > across multiple function calls. > >Suppose multiple processes open the same file with optuon O_APPEND. > >That option requires that each write be done at the end of file > >at that tine, not at a "retain[ed] position." > >This requires a clarification. -- gil ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
