Code Blit:

The string module has translate and maketrans, through multiple versions. In per 3.x,

an optional argument listed "letters to delete".

The first argument was a kind of mapping, of 256 symbols to itself, except with some

letters changed, what we would call a "permutation" in group theory (not promising

a bijectional one-to-one mapping, no way, more likely surjective, do we agree?).

Here we are in 2.6, where the bytes type is not yet operative.

IDLE 2.6.5

>>> b'a'

'a'

>>> type(b'a')

Now let's replay that action in 3.1:

>>> b'a'

b'a'

>>> type(b'a')

Here's my idea of a punctuation filter in 3.x, where the delete characters clause

is now a delete bytes clause in the byte type superobject.

import string

def filterpunc(thestr):

return thestr.encode('utf-8').translate(bytes.maketrans(b'',b''), string.punctuation.encode('utf-8')).decode('utf-8')

See:

if curious.

This'd be used to strip string.punctuation characters out of any string. Other solutions possible, including with regexps.

Suggestions for Features (SFFs -- not as prestigious or important as PEPs

but sometimes the genetic precursor of a PEP).

Save Session in IDLE or other PyShell:

This is probably already out there, in IPython or one of those. It's

so cool to be able to interactively bind names to working code in

memory, but when you resume your session later, all that has

gone away. In some other shell environments, you have the

option to hibernate and resume, Smalltalk's being the most famous.

Would it come with a source code snapshot version that encapsulates

all the definitions as runnable code? The latest version of f, all

the latest object values? That's another way to "pickle", to just

write out a module that recreates what was there, when run.

Python in Mathland:

A link to a recent summary that mentions Python quite a bit:

Kirby