A little birdy pointed me to this discussion (note the three linked
articles):
http://fairhaven.typepad.com/my_weblog/2008/11/the-unixntp-real-world-compromise.html
This is a brief, but reasonable, analysis of broad timekeeping
requirements for some healthcare scenarios.
A few comments:
1) Whatever the future status of international timekeeping standards,
such analyses are required for pragmatic deployment of innumerable
systems worldwide. In particular, I like the author's discussion of
scope (admin vs operating room, for instance).
1a) Eradicating leap seconds won't address most timekeeping issues in
the least.
2) Requirements for an international standard should logically trace
to requirements for the full ensemble of timekeeping users. This
ensemble is large (very) and diverse (very).
3) Note that they speculate on an acceptable solution, a variation on
one of the timeslicing schemes that was discussed very early on this
list. The main point here is that the acceptability of a solution
depends on the context. (As a patient, though, I may have concerns
about their timekeeping choices :-)
4) Any project is required to deal with standards as they are. A
standards effort, on the other hand, is about finding a satisfactory
solution for a (weighted) average of all projects. A lot of our
disputes are really about asserting an acceptable weighting function.
That is, one position is that astronomical applications get a zero
weighting because astronomers are - well - able to overcome bad
standards, while (for instance) consumer electronics applications get
a 100% weighting because corporate engineers are (willfully) ignorant
of timekeeping issues.
I might assert that a successful strategy here would involve a
somewhat more nuanced derivation of requirements. In particular,
ignorance of the issues also leaves one ignorant of potential hidden
dependencies in the requirements.
(With the economy the way it is, one is also skeptical of corporate
style decision-making at all levels.)
Rob Seaman
NOAO
_______________________________________________
LEAPSECS mailing list
LEAPSECS@leapsecond.com
http://six.pairlist.net/mailman/listinfo/leapsecs
