On Wed, 28 Oct 2015 08:10:43 -0400, Peter Relson wrote:
>
>I suspect that, if you asked your colleagues in an impartial way, they
>would not agree with your opinion on this topic. To my recollection you
>have not initiated such a discussion, but you do continue to post
>statements like this. It is true that IBM refuses to document anything
>about local, GMT, or UTC time for these services because the services are
>nothing more than (architected-format) clock value manipulators, which is
>what is documented. Make your case.
> 
I don't know the chronology, so this is somewhat conjectural.

I believe that prior to the advent of leap seconds in 1972, "clock value
manipulators" (not documented as well as you'd like to believe) sufficed
to convert between TOD values and GMT.  Progrmmers could use STCKCONV
to convert timestamps, even archival, to displayable values.  I belive this
was the primary design objective of STCKCONV and CONVTOD.

In 1972 leap seconds broke this ability.  STCKCONV was not updated to
account for this innovation.

Assembler Services Reference says, "STCKCONV time of day and date formats are
compatible with the formats returned by the TIME macro ..."  It does not say 
that
STCK followed by STCKCONV gives the same value as TIME would have given at
the instant of the STCK, but a programmer might be misled, with wishful 
thinking,
to believe so.  A sentence of caution in the Reference would be in order.  
Better,
an example showing how STCK, then subtracting CVTLSO or a value from a table
in the Principles of Operation could be used to construct a displayable time 
value.

This is a good forum for the discussion you invite.  How do readers use 
STCKCONV?
To convert current STCK values?  Why not simply use the TIME macro?  For
archival STCK values?  Do you apply offsets from the Principles, or simply 
regard
the LSO, currently almost a half minute, as inconsequential?

-- gil

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to