On Tue, 8 May 2007 20:52:06 -0500, McKown, John wrote: >> >> DEFINE ALIAS( NAME( user.WHATEVER ) - >> SYMBOLICRELATE( user.WHATEVER.D&YYMMDD ) ) >> >> But, doesn't this mostly dispel the rationale repeatedly >> given for not allowing references to &YYMMDD in batch JCL, >> viz that earlier processes (Reader? Converter? Other?) need >> to know how system symbols will be resolved to build real >> data set names? > >I don't think so. The date at which the DEFINE occurs and the system >upon which it occurs is known at the time of the DEFINE. With JCL, the > But you misunderstand how it works. The symbol is bound to its value not at the time of the DEFINE, but when the data set is accessed. For example, a few weeks ago as a test I defined such an alias. Immediately afterward, it referred to data set 'user.WHATEVER.D070420', the then current date. Today, it refers to 'user.WHATEVER.D070509'.
>concept of "date" is generally ill defined and could be argued for many >interpretations. Such as: (1) Reader Start Date; (2) Job start date; (3) >step start date; or perhaps even (4) date when the dataset was first >OPENed by the step in which it was defined. Personally, I like (1), but >others might argue for the other possibilities. In the case of (2), what >about for a job which uses z/OS step restart? Which job start date? The >original job start date or the date upon which the job was restarted? Or >perhaps even worse would be if someone wanted to "integrate" this with a >product such as CA-11. Again, which job start date? Or even which Reader >Start date? An STC does not have this particular problem. > So, if I had run my job a few hours ago, on the cusp of midnight, would it have accessed via the symbolic alias a data set name incorporating (1) Reader Start Date; (2) Job start date; (3) step start date; or perhaps even (4) date when the dataset was first OPENed? And, more to the point, if such uncertainty is permissible for data set names accessed via symbolic aliases, why is it not likewise tolerated in data set names directly incorporating system symbols? >> And how would the JES3 scheduler deal with such a construct? -- 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
