On Tue, 30 Oct 2007 10:19:46 -0500, McKown, John wrote: > >> Indeed, but too little too late. The OS (not utilities) simply >> should _never_ change the attributes of an existing nonempty >> data set. ... > >I'll go futher than that. The entire concept of LRECL, BLKSIZE, and >RECFM is archaic and should be eliminated. If a program wants to read aQ >dataset, let it specify what it thinks the LRECL and RECFM should be. If >the system can accomodate that, then so be it - let the system (access >method) read the physical data and present it in the format that the > Quite as the BSAM/QSAM interface to HFS/NFS/ZFS operates. The implementors of that facility were clever. Much the same for PDSE wrt BLKSIZE, though not LRECL nor RECFM.
Wouldn't it be nice to have RECFM=STREAM,LINEND=NL, in addition? (Would probably work best on FBA DASD.) >program wants. The only cavaet is if there is a record which cannot be >processed properly if "reformatted" into the LRECL that the program says >that it can accept. > UNIX FS interface gives I/O error: "READ WRONG LENGTH RECORD" for this. At its glacial pace, IBM is accommodating twentieth century technology. -- gil ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
