On Thu, May 19, 2011 at 11:13 AM, Denenberg, Ray <[log in to unmask]> wrote: > Please may I have opinions on whether these use cases justify the additional complexity, particularly from those of you who have commented that the feature adds too much complexity. I'm really concerned about a workable spec: one that is likely to see wide support not only in a variety of data formats, but also in related software libraries, UIs, etc. I originally supported questionable dates, but I do not support adding them to a level of detail that conflicts with the above goals. So, I would support a single trailing "?" that applies to a valid ISO date representation*, but think allowing it to be applied to any component within that goes too far because it fundamentally changes ISO dates (not sure I put that well, but hope you get the idea). Bruce * e.g., any of these seem reasonable to me: 2000? (an uncertain year) 2000-03? (an uncertain year-month) 2000-03-04? (an uncertain year-month-day) 2000-04?/2003-05 (a range with an uncertain begin year-month) ... while these do not: 2000?-03 2000?-03-04? 2000?-(03-04)?