Date: Sun, 6 Jan 2013 18:29:04 +0000 From: Shawn Steele <Shawn.Steele@microsoft.com> Message-ID: <a1041674394d4a9f98ddb57f08a6e382@BN1PR03MB040.namprd03.prod.outlook.com> | but are there *any* known cases where people make exceptions to | the standard? That depends, I guess, on what it would take to qualify as a case that counts for your purposes. I have certainly created new timezones when I have been testing stuff, kind of "what if" cases for the timezone database, and/or for making sure that my understanding of how things work matches reality. But none of those zones ever get used for anything "real", no ordinary application would ever see them, and I certainly have no need for any standardised kind of name for them (TZ=$PWD/test1 (or something) works just fine) In general, we attempt to cover all timezones that anyone could ever want - as we attempt to make sure the database contains everything that exists (for many people, far more than they claim to need, as many people, at least claim, not to care about historical differences). That makes it rare for anyone to need to create a new real zone - if it is important enough for someone to need to do that, then it probably qualifies to be included in the tz database. | Even if they're reusing names? Reusing names would just be altering the definition of a zone, which, other than to update it, or correct an error, doesn't seem like a particularly useful thing to do (we have no shortage of file names, creating a new one is not a difficult task...) | (Reusing names for different behavior seems bad for interop). There isn't really any interop issue, at least, not for the names as they're defined and used here, they're purely a system local identifier that selects a particular timezone - we don't expect them to be passed around (other groups define timezones that are used for interoperability purposes, like shared calendars, and I think have different naming schemes, as much as names are needed at all.) kre