Issues w/Daylight Savings Time in 2007 ?

[Werner Koch]

On Thu, 28 Dec 2006 07:19, rjh at sixdemonbag.org said:

> Well, before the changeover happens, I'd appreciate it if a switch would
> be added to preserve the current behavior.  Seconds since epoch is dead
> simple and very familiar to programmers, and I suspect it might be
> better for machine parsing.

I have no concrete plans for that change.  With gpgsm we use an iso
based format for all internal timestamps because that is the only way
to get around the year 2038 problem as well as some other minor
problems.  For X.509 this is required because there are real world
certificates valid for more than 30 years (I am not sure whether "real
world" is considered a joke for some CAs).

For OpenPGP the protocol uses seconds since epoch and internally we
handle it all as an unsigned 32 bit type so that we are protocol-wise
safe for another 100 years.  Given that it is unlikely that on GNU
based systems time_t will be changed to 64 bit any time soon, we need
to do some internal cleanups in a few years and eventually change the
output format of thye --with-colon listing.  Of course, there will be
an option to enable a new format. 

FWIW, gpgme automagically detects whether we are using seconds since
Epoch or ISO time.  New programs using gpg directly might want to do
that additional check (and conversion) too to be prepared for the
future.


Shalom-Salam,

   Werner