Fwd: A structured format for dates?

[iain hill <[e-mail]>]

Oops, I meant to send this to the group, sorry for the duplicate 
Poul-Henning Kamp.


~~~

Poul-Henning Kamp writes:

> Time-nut-PHK would vote for such a proposal, but programmer-PHK would
> be firmly set against it.

If I might be so bold, HTTP is specifically used in conjunction with the 
REST thesis and consiquent paradigm, which is itself clear predisposed 
for use in very large distributed systems, as the web very much is.

I would agree that any local program can and does benefit from having a 
very precise 'local time', however any programmer of distributed 
systems, ought be well aware of the advantages of using a dynamic time 
type rather than a rigid one, as it is the only way to synchronize such 
a system.

> When it comes to computers and timekeeping, "doing it right" has not
> been an option for decades.
Babbage would be horrified; His goal was to remove human introduced 
errors from ephemera calculations.
> Given my painfully extensive experience with my own and other peoples
> time-bugs, I am now firmly in the "Call time(2), use that number,
> as a number, but dont modify it!" camp.

I seems to be the case that this subject need to be thought about far 
more carefully when educating the programming discipline. That said, 
there are still issues in comprehending this subject in the field of 
Astronomy also, so we are far from resting at the inn as yet!

We might consider that the very same issue is also very apparent in race 
conditions, within concurrent code; Seeming somehow to be steeped in 
mystery as a bug, yet in reality it is the exact same problem on a 
different scale.

We are dealing with a distributed system, not any one programmers local 
machine; Acknowledging the existence of different time types seems to me 
to be a fundamental requirement for any system, that is to be scaled, 
over any consequent space and that maintains or uses multiple runtimes.

It would seem to me that we might reasonably conclude that we are very 
far from lucidity on matters of time right now across multiple domains.

Kind regards.

Iain