On Tue, 10 Jan 2012 10:23:37 -0600, Paul Gilmartin wrote: >Let me give an example. Suppose after I have APPLYed PTFs >A, B, and C in sequence I detect a bug. I'd like to isolate the >causing PTF. So I do what is necessary to RESTORE C and >test again. The bug is still there. So I'd like to RESTORE B >and test yet again. But I can't because in order to RESTORE C >I had to ACCEPT B, and now it can't be RESTOREd. This >is terrible; it's a deficiency in design.
You didn't have to ACCEPT B or A. Indeed, in the example that you gave, it would be foolish to ACCEPT A or B. What you should have done in that instance, assuming that A, B and C all modified some of the same elements, is to RESTORE A, B and C, then apply A and B. >Your assertion, "That is not what SMP/E does," is not a refutation >of Robert's complaints and mine, but a confirmation that it >fails to support needed function. I made no comments about your or Mr. Rosenberg's complaints about the design of SMP/E. My comments were about his incorrect description of what SMP/E does during RESTORE processing. -- Tom Marchant ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
