On Fri, 29 Mar 2013 09:08:36 -0400, Peter Relson wrote: >>I did not find Kevin Kelley's post entirely persuasive. This >>restriction long antedates 2 Kibyte pages, and the equation 8 x 4096 = >>32768 is thus historically irrelevant. > >Kevin was not trying to persuade. He was merely stating facts. The number >of pages is not the relevant part of the information. The relevant >limitation is the number of bytes that a halfword can support. > Don't attach too much significance to Gilmorespeak. I took it that he merely was not persuaded of the validity of the argument.
>Aside from the BLKSIZE point, we would not consider a number about 32767 >(for example 65528) as the limit, for compatibility reasons. There are >surely programs that use "LH" on the 2-byte length field without >subsequently clearing the high bytes and then rely on the value that would >misbehave if they land with a negative number. > A horrid example is HLASM. I tried it. Dog in the manger. Similar arguments could be made (and were in May, 2005) against PARM > 100 (overrun of traditional sized buffers) and PARM >256 (misbehavior of MVC). Fortunately, z/OS Core Technology Design did not find them persuasive when considering the PARM enhancement for z/OS 2.1. -- gil ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
