> I have always considered "timestamp with time zone" to be a bad > description of that data type but it appears to be a carryover > from the specs. It is really a "point in time"
I agree. While what timestamptz implements is a very useful data type, I think it was a very unfortunate decision to implement that for the standard type name, instead of something more consistent with the spec. It seems very unlikely to change, though, because so much existing production code would break. :-(
Understandably, people do tend to expect that saving something into a column defined as TIMESTAMP WITH TIME ZONE will save a time zone with the timestamp, and in PostgreSQL it does not.
Yes, it strange for first moment, and it is difficult for beginners - but it works well .. after you switch to different mode.
But can we implement a Time Zone as special type? This and examples and documentation can better explain what it does.