On Thu, 10 Dec 2015 15:29:24 -0500, Robert A. Rosenberg wrote: >> >>QNAME SPFEDIT is only used for PDS[E]. SYSDSN is another issue, but >>there the volser is not known in time to include in the RNAME. > >I agree about SYSDSN being issued at times (JCL DISP for example) >where the VOLSER may not yet be known so it can not be used as part >of the RNAME. The fact that there are times when the VOLSER *IS* >known does not help due to the unknown VOLSER cases. I think that >this situation is one of the things that JES3 handles by not letting >a JOB be submitted/run unless the DSN is not being referenced by any >other job's JCL. > But this becmes increasingly inappropriate with increasing use of DYNALLOC by TSO/E, ISPF, FTP, NFS, etc.
It appears that NFS (and FTP?) use SPFEDIT-style PDS serialization. I understand that PDSE supports concurrent writing to different members of the same PDSE. (What scope? Same task? Different address spaces? Different LPARs?) A while ago, I did a crude experiment which appeared to show that ISPF is not savvy to this capablity: LMINIT for Write two different members of the same PDSE and perform interleaved LMPUTs to them. Failed somewhere along the way. I might try this with NFS. Would ISPF-L be a better forum for this topic? -- gil ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
