2001 Enchancement Wishlist

[Alex Martelli]

"Raymond Hettinger" <othello at javanet.com> wrote in message
news:3A4CEACB.220998A0 at javanet.com...
> If these ideas meet with favor, I'll write-up a PEP.
>
> 1.  List.append(elem) and List.pop() are great.  Let's add two more
>      methods, List.prepend(elem) and List.leftpop().  The goal is to

Pass (apart from the unhappy names); I think .insert, and slice
assignment in general, are by far enough, but it's a min.r.

> 2. There is already a PEP to extend the eval() function which is currently
>     limited to dictionaries.  The goal is to accept any object defining
>     __getitem__() so that any mapping object can be used to specify the
>    globals and locals.   I recommend that we extend the same courtesy to
>    the % string formatting operator.  For example:

Sure, the more polymorphism, the merrier.  That one I'd like to see.

> 3.  __init__ is the only special method that doesn't allow a return value.

I think it's better off without one, just as it is now.  Use factory
functions if you want to generate with more freedom, and let
class objects as callable make-and-initialize new instances.

> 4.  Allow dictionaries to respond to sequence operations as if .keys()
were
>      in the list object.  Examples:
>
>     knights = { 'lancalot':'good', 'gallahad':'brave', 'robin':'coward' }
    [snip]
>     print knights[0]              # may be any one depending on the hash
table

Unfeasible.  0 (or any other int) might be a key (and it may happen to
fall anywhere in the hash table order, too).  So what does knights[x]
mean -- the value corresponding to key x is present, as now; or, the
value corresponding to key x is present, or else if x is (or, can be
coerced to?) int, the x-th key in hashtable order?  The latter is far
too complex -- and all the complexity for very little gain.

This one would be a real blooper if Python adopted it, IMHO.


Alex