On Sun, 7 Oct 2012 17:10:16 -0400, Gerhard Postpischil wrote: > >I just ran a series of tests, writing a VB data set, then a VBS >data set, with various data lengths from 0 to 20, once with move >mode and once with locate. All ran fine. Then I ran analogous >tests reading the data sets, and dumping the input; an >additional read with BFTEK=A also worked correctly. > >So the description doesn't match QSAM behavior; is it possible >the author was thinking of compiler library code? > It's possible that when the specification for RECFM=VB was changed, the doc writer overlooked the corresponding change for RECFM=VBS. The doc and the product ought to be brought into agreement, either by changing the doc or reporting the error.
But this is all empirical. It's possible that a programmer writing a utility for IBM or ISV uses BSAM and conscientiously checks for all errors and reports the occurrence of an empty segment in a VBS data set, shich is contrary to spec (do you agree?) as an error. If it's contrary to spec and it breaks you're allowed to keep both pieces. Does BFTEK=A apply to output, or only to input? If an empty segment is allowed as an interior segment, a programmer could inflate a logical record with an arbitrary number of empty segments. Ugh! -- gil ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
