[LEAPSECS] EBML: yet another date format?

[Daniel R. Tobias]

It appears that yet another date format (queue up that XKCD comic on 
standards...) is being proposed as part of the spec for EBML, a 
binary markup language intended to be a binary-file equivalent of 
XML. The way dates are defined there is:

     signed 8 octets integer in nanoseconds with 0
	indicating the precise beginning of the
	millennium (at 2001-01-01T00:00:00.000000000 UTC)

No indication is given of how leap seconds are supposed to be 
handled; if this is to be a continuous count, then converting it to 
human-readable dates/times would require knowing the complete table 
of leap seconds (and knowing what time scale is to be used 
proleptically if dealing with dates prior to 1972). Or if it's to be 
treated like Unix-time dates where leap seconds are ignored, then 
there wouldn't be any way to represent the times within the leap 
second. As long as more specifics are not given on these points, 
implementers are bound to make contradictory assumptions and get this 
standard as screwed-up as many existing ones are.

Can somebody do the math to figure out what range of dates would be 
supported by a signed 8-octet integer in nanoseconds centered on 
2001-01-01?


-- 
== Dan ==
Dan's Mail Format Site: http://mailformat.dan.info/
Dan's Web Tips: http://webtips.dan.info/
Dan's Domain Site: http://domains.dan.info/