I gave a presentation on Stored Processes back in 1989 in Johannesburg, to SUGISA (SAS Users Group In South Africa).
Back when we had slow processors, compiling the code of a large datastep took longer than processing the data. In my case I reduced a monthly report run from 40 minutes to just 2 using the above. This was one datastep of over 3000 lines churning through two datasets of over 500 variables each - a legacy of using FSEdit on the mainframe and slow response times (remember those?). I never even thought that this knowledge would ever be used again, but it seems that is now not the case.
Anyone else have any tales from the past to share?