On 2014-06-23, at 07:50, Peter Relson wrote:
>> Do they deal in LOCAL time? GMT? TAI? Other?
>
> The reason that no "clarification" to STCKCONV/CONVTOD was made was
> because no clarification seemed appropriate (and I continue to believe
> that). These services deal with the TOD clock. In my mind, a TOD clock
> value is not local, gmt, tai or anything else other than itself. The
> important concepts to the TOD clock are that 1 microsecond is represented
> by bit 51 and what a value of 0 represents.
>
> After you have dealt with the clock value, you can then talk about
> converting it to another format.
>
> You could say that the output from STCKCONV and the input to CONVTOD is a
> date/time according to TOD clock rules. Perhaps I'm too close to it, but
> given that the input (to STCKCONV) or the output (from CONVTOD) is in TOD
> clock format, I think that such a statement would be somewhat redundant.
>
Once you mention "another format" further specification of that other
format is needed. For example, for the TIME macro:
http://pic.dhe.ibm.com/infocenter/zos/v2r1/topic/com.ibm.zos.v2r1.ieaa900/iea3a9_Description20.htm
Description
z/OS MVS Programming: Assembler Services Reference IAR-XCT
SA23-1370-00
The TIME macro returns the local time of day and date, the Coordinated
Universal Time (UTC) (or the Greenwich mean time) of day and date,
"local" or "UTC" simple enough.
The section goes on to describe STCKCONV and CONVTOD, readily leading the
reader to believe that STCKCONV returns the value that TIME would have
returned had it been invoked when the STCK was performed, and that CONVTOD
returns the value that STCK would have returned at the time TIME was
invoked.
That would be ideal, but I suspect it's not the case, and that the formatted
time represents
GMT for TOD values prior to 1972
TAI - 10 seconds for 1972 and afterwards.
(I'd be delighted to be proven wrong.)
An alternative OS provides a more complete specification; incomplete
and flawed, but they tried:
http://pubs.opengroup.org/onlinepubs/9699919799/basedefs/V1_chap04.html#tag_04_15
A citation of the PoOp would suffice if it enabled the reader with
paper, pencil, and persistence to replicate or verify the behavior
of STCKCONV and CONVTOD.
-- gil
