Michael, I wanted to start an issue to track session.truncate.

I went ahead and tried to fix up the documentation, it now reads that the
URI and the stop/start cursors are mutually exclusive (so, you're either
truncating a table or you're discarding between two points in the table),
and I noted that in the latter case the start/stop points are inclusive.

I also went ahead and implemented the easy parts:

I think session.truncate on a URI works: I used the schema_drop code as a
template and simply truncated the files and discarded their root addresses
from the schema file).

For start-to-mid-point and mid-point-to-start cursor truncation, I used the
obvious loop. We could optimize this if we wanted to by reaching down
into the files and doing something faster, but I figured for 1.0 we didn't
need to get that creative.

I don't know how to do the missing part, which is compare two table-level
cursors, which is why I've assigned this Issue to you.

The code I did needs review, just to make sure I didn't get some schema
modification code wrong.

Thanks,
--keith

=========================
Keith Bostic
keith.bostic@wiredtiger.com