Kurt Zeilenga writes: RFC 4517 says: The "Z" form of <g-time-zone> SHOULD be used in preference to <g-differential>.
The primary reason for this is interoperability. Not all implementations can deal (well) with g-differential. It should be avoided (on the wire).
Well, I'm not that eager for it.
Though I don't see that as much of a problem - in the sense that if the admin asks for it, it's his responsibility to deal with it. We've got other non-interoperability options too. Like {crypt} in userPassword, which even breaks the standard.
Owen DeLong writes:
Another consideration is that g-differential may be different in a given location, depending on the date.
For example, in California, we are -0800 during the fall/winter and -0700 during the spring/summer due to daylight savings silliness.
Which but the software only needs to know for its own host, and localtime() + strftime("%z & co") is supposed to take care of that.