fix: enhancement to history command output and % template fix. Rather than using the key field, use the label field for descriptions. Call cls.labelprop(default_to_id=True) so it returns id rather than the first sorted property name. If labelprop() returns 'id' or 'title', we return nothing. 'id' means there is no label set and no properties named 'name' or 'title'. So have the caller do whatever it wants (prepend classname for example) when there is no human readable name. This prevents %(name)s%(key)s from producing: 23(23). Also don't accept the 'title' property. Titles can be too long. Arguably we could: '%(name)20s' to limit the title length. However without ellipses or something truncating the title might be confusing. So again pretend there is no human readable name.

# In Python 3 the "anydbm" module was renamed to be "dbm" which is now a
# package containing the various implementations. The "wichdb" module's
# whichdb() function was moved to the new "dbm" module.

try:
    # old school first because <3 had a "dbm" module too...
    import anydbm
    from whichdb import whichdb
except ImportError:
    # python 3+
    import dbm as anydbm
    whichdb = anydbm.whichdb