> Perhaps "---", since we often use "-" to indicate missing > stuff. But pretty much anything that is an "out of badn", > e.g. not a digit, would be acceptable.
Unfortunately, this would violate [RFC3339]. I think we should try to stick with the accepted standard rather then inventing yet another time stamp. I think it is easy enough to parse RFC3339, the variable length should not be a to-big issue. Especially as the token is well-defined, so there is no doubt where it will start and end. This is different to the tag value that Andrew used as a sample (well, it is defined in [RFC3164], but in reality there are a number of variations and "smart detection" causes lots of trouble. Andrew: can you agree on this? Anybody else oposing it? Rainer
