> I would like to see a SHARE requirement that PDSE's should be usable > everywhere a PDS is including IPL time LPA and LINKLIST data sets, > PARMLIB datasets and SYS1.NUCLEUS unless that is replaced by an IPL > set of cylinders (who would care if IPLTEXT were 800 or 1000 cylinders > provided we were given information clearly and in advance?). This > might motivate IBM to fix the PDSEs. Note, I also believe a variant > of the PDSE should support long (at least 63 character) member names > in UNICODE). We aren't in the world of 28 megabyte 2314s any longer.
All good points. While I happen to be a supporter and user of PDSE, I won't make any effort to defend how tragically screwed up and flat-out buggy that technology has been from the beginning. And there is literally NO excuse for the fact that you can't use them for IPL-time system functions or that so many system utilities don't support PDSE today more than a decade after they were introduced. It is not that such changes can't be made. The (fairly trivial) removal of the SMS requirement proved fairly convincingly that many of the heretofore "restrictions" were and are artificial. You can open and read a PDSE very early in the IPL and contents supervision handles them just fine. Been there and done that in shipping code! But most of the PDSE support code is (I believe) still in the LPA, so you can't do anything with a PDSE before the LPA is built, which is after the linklib is built... etc. I guess there's an analogy about carts and horses here somewhere. Repackaging the relevant support functions so they could be used earlier is technically feasible, but it is likely to be estimated as "a lot of work" in terms of release line items. Changing all of the other system utilities to handle PDSE datasets is an even bigger and more complicated job. Other new function requirements would almost always trump this scale of change to a single component, even if that component is important. So much for the "new PDS" idea. Now if the changes you're suggesting were wrapped up in some larger set of functional requirements you might be in with a shot. But I will bet you a buck it ain't going to happen. CC ---------------------------------------------------------------------- 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
