RKOGUT@BROWNVM.BITNET ("R. Kogut") (06/16/89)
I know of no one else other than PL/I that actually uses dummy sections, but the idea is twofold. One is to provide a common data area that can be accessed by all subroutine, assuming that the address of the data area is always passed as a parameter or otherwise accessible. A-cons and the like aren't very useful for programs invoked with LOAD, LINK, and XCTL, for example. The second apsect is the ability to tailor the actual data area to the options, features, components, or whatever of the implementation actually being used, and to insulate against changes. For example, suppose a DCB for an audit file was placed in a data area, and the size of the DCB was expanded in a later release of OS. Simply recompiling the routine that allocated the common area would be sufficient, as opposed to all of the components.
DEEHRE01@ULKYVM.BITNET ("Dr. David Ehresman") (06/17/89)
*** Reply to note of 06/16/89 09:44 This is a test reply. Please ignore and forgive me.