[Types-sig] Suggestions for python 2

Terry Reedy tjreedy at udel.edu
Sun Jan 16 22:03:26 EST 2000


Initial responses after trying to read the web-page proposal:

The tunneling idea of allowing name=expression pairs after *,** in function
definitions to indicate read-only compile-time-evaluated local constants
seems like a good one in that it would eliminate the problem with the
current pre-* default param hack (modifiability) without disabling current
correct code.

Much of the rest I did not get too well (but I feel better after reading
Guido's post).  One of the benefits of the current Python object system is
that it is fairly clear and simple.  This means that non-experts without a
CS degree can understand and use it.  It also means that such people
(including myself) can read and understand most code written by others.  The
straightjack which understandably frustrates some CS-level experts has some
benefit for the rest of us.

Some unification of similar concepts might be good, but not at the cost of
what seems like substantial complexification.

Terry J. Reedy






More information about the Python-list mailing list