In email, keith.bostic wrote in reply to sulabh.mahajan:
On Sun, Sep 10, 2017 at 9:04 PM, Sulabh Mahajan <sulabh.mahajan@10gen.com> wrote: > The semantic says setting the oldest timestamp older than the current is a no-op Sulabh, as far as I can tell, the documentation doesn't discuss these kinds of exceptions and/or errors (for example, WT_CONNECTION.set_timestamp currently says "Supplied values must be monotonically increasing"). Would it be possible for you to review/fix the docs based on your work?
And Sulabh agree to review the documentation. I have found other inconsistencies in timestamp documentation that should be addressed at the same time.