It'd be just about as legitimate as Gadafi's rule is.

It'd be just about as legitimate as Gadafi's rule is.

$ date -d 'November 16, 1999 PST' +%s
942739200
$ date -d @942739200
Tue Nov 16 00:00:00 PST 1999

It tracks all of the changes. If a train left Paris at 12:06pm on Jan 12, 1936 and averaged 60mph all the way to Moscow, you can't tell when it arrives in Moscow just by knowing the timezones now and the length of the track. You need to know what the timezones were at the time. And, you need to know exactly when they changed -- especially if they changed during the journey.

Also, since zoneinfo doesn't guarantee integrity for data before 1970 and Unix time can't even express it, that's not even a good example! :)

The database is designed in a smart enough way that those problems are just a series of easy steps that can be abstracted away from the perspective of the C libraries in question. Replace "complicated" with "elegantly modularized according to Unix tradition."

It's a surprise the C library that uses these files can manage to keep all the time accounting straight...

It's not that complicated. They all translate into offsets in seconds. To the computer, I don't live in America/Los_Angeles on 3:47pm Thu March 3, 2011. The computer sees:

1299196020 (unix time in UTC)
- 28800 (my zone offset in seconds, using the tz database)
+ 0 (no DST in my zone right now)
= 1299167220 (local time)

So the really impressive work has just been in conceptualizing and organizing the database so that a program just needs to lookup two questions: which of the zones am I in, and what is the current offset for that zone?

The tz database is meant to keep track of present and historical changes. Your proposed changes would not simplify anything - they'd only make the tz database bigger.

This. There are always new problems to be solved. Instead of complaining about how easy it is to do the same things we did in 1995, how about dedicating all your free time and ostensible baremetal skills to writing patches that solve TODAY's problems and tomorrow's problems? Help with btrfs, Tor, SystemD, DNSsec, or how about writing the code to finally make xinerama with compositing work again in xorg? That is currently an OPEN BUG. Anyone who says they miss the grit, and isn't spending their time on this stuff, is just all talk.

"The most important difference from conventional hacking is that the fraudsters do not need to log into the online banking computers - they simply ride on the existing and authenticated session, much as a child might slip in unnoticed through a turnstile at a sports event, train station, etc."

Like putting too much air in a balloon!

sudo make me a sandwich

My digital alarm clock already invented that.

That's what it sounds like to me.