[LEAPSECS] Longer horizon

[Warner Losh]

On Jul 10, 2012, at 11:17 AM, Rob Seaman wrote:

> Your message seems snarkier (more "cranky, irritable") than mine.  You speculate on what I do or don't understand, and on what I am or am not doing.  All of these are irrelevant.  I'm a big fan of FreeBSD and PHK's MD5 password hashing, but still disagree with his position on leap seconds :-)


Basically, the problems are deeper seeded than what a simple SIGHUP would fix.  They are deeply engrained in the culture of the software world, and failing to acknowledge that fact, or being dismissive about how "easy" it is to implement doesn't help.  That's the point I was trying to make.  Sorry if I was too grumpy in making it.


> There has been only one proposal on the table for thirteen years.  The proposal is unacceptable.  Its backers do not participate here, where we have discussed numerous alternatives.  It isn't the astronomers (themselves mostly software professionals) who have been unwilling to consider the options.  If this is a software issue it would help if software solutions were permitted to be discussed, and not be rejected out of hand.  That said, the ITU proposal has precious little to do with software.


I've been trying to propose other things that would make things work better most of that time.  That's where the 'tell the world about it sooner' thread has come from.  The longer that you know in advance, the better things are all around.  Leap seconds aren't just a time zone thing, they wind up impacting more things than you might naively thing had you not implemented a system that tried to get them right.  It also helps you understand the consequences of missing knowledge at different parts in the system that you might think is always there, but turns out to not be in a number of interesting cases.

Warner