Hello! > I suggest that we not get > sidetracked into engineering > a solution, until someone > speaks up and claims that > this is a requirement. I agree and don't aim at engineering a solution. Some clarifications about the "philosophy" behind the specs may be useful, though. > > uuuu-uu-uu/1760-12-03 > [...] Is it an interval or a specific date? Is it obvious how long time something needs to take to be considered taking an interval of time? > > (replacing "2011-03-10" by the day > > when the document came to the library or > > archive, etc. or another appropriate day) > That's a completely separate use case I wrote "or another appropriate day", meaning a day which for sure was after the end. This was of course never intended to be a primary way to store the acquisition date. (read below) > 20040101/ Does anyone see a problem I consider that the notation is OK. But, are such open-ended intervals really useful? My aim is to avoid what I call "self-destructivity of stored information", that is, information stored in such a way that it looses its accuracy with time (no matter the specific notation used). I quote: > A syntax like ".ea.1760-12-03" contains > an uncertainity growing with time. In one > year or 500 years, the reader of the > notation ".ea.1760-12-03" will be given a > larger incertainity than a reader is > given today. Expressed an other way: The > quality (or accuracy) of the information > stored through the notation > ".ea.1760-12-03" is not preserved as time > goes. I consider that preservation is one > of the main goals for libraries, museums, > archives etc. I consider that closing the "end side" of the interval protects the information from further loss of accuracy. Regards! SaaĊĦha,