On Fri, 31 Oct 2008 18:53:12 +0000, john gilmore wrote:
>I have read the sequence of contributions to this thread with more dismay than
>usual. Working with leap seconds poses no particular difficulties, and in
>particular Paul Gilmartin's suggestions are otiose. (I single him out only
>because he almost always knows what he is talking about, even when I disagree
>strongly with him.)
>
Otiose? I'd prefer to be seen not as spearhead of the campaign
to abolish leap seconds but as reporter and commentator.
What more than a sinistral compliment can I expect on Hallowe'en?
>A table is needed but IBM supplies one, which it updates well in advance of
>the effective insertion time of a new leap second, indeed almost as soon as
>the BIPM issues its notification that an insertion will occur. ...
>Suppose now that we have a copy of this table, call it T, and a (truncated)
>STCKE value, call it S, for which we need the corresponding cumulative
>leap-seconds-inserted value. Traditional match-seeking binary search in this
>table is unavailable to us because we need not a match for but a greatest
>lower bound (glb), which may occasionally also be a match, on S. A
>binary-search scheme that I devised long ago for evaluating tabular step
>functions can, however, be used for this purpose.
>
I believe we agree that it's practical; "not rocket science".
I'd be delighted to see IBM incorporate such an algorithm in
the implementation of the STCKCONV macro and update it promptly
by PTF upon the notification that an insertion will occur.
I'm far from delighted to see it left to individual sites,
programmers, and ISVs, all of whom are less likely to perform
timely updates of their code.
And I'd like to see a similar treatment of the chronicle
of Daylight Saving Time changes as many (most?) suppliers
other than IBM provide:
Linkname: Zoneinfo - Wikipedia, the free encyclopedia
URL: http://en.wikipedia.org/wiki/Zoneinfo
Linkname: The zoneinfo home page
URL: http://www.twinsun.com/tz/tz-link.htm
-- gil
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
