Common LISP-style closures with Python

[Antti J Ylikoski]

In Python textbooks that I have read, it is usually not mentioned that
we can very easily program Common LISP-style closures with Python. It
is done as follows:

-------------------------------------

# Make a Common LISP-like closure with Python.
#
# Antti J Ylikoski 02-03-2012.

def f1():
n = 0
def f2():
nonlocal n
n += 1
return n
return f2

-------------------------------------

and now we can do:

-------------------------------------

-------------------------------------

i. e. we can have several functions with private local states which
are kept between function calls, in other words we can have Common
LISP-like closures.

yours, Antti J Ylikoski
Helsinki, Finland, the EU

 

[Chris Rebert]

In Python textbooks that I have read, it is usually not mentioned that
we can very easily program Common LISP-style closures with Python.  It
is done as follows:

-------------------------------------

# Make a Common LISP-like closure with Python.
#
# Antti J Ylikoski 02-03-2012.

def f1():
   n = 0
   def f2():
       nonlocal n
       n += 1
       return n
   return f2

i. e. we can have several functions with private local states which
are kept between function calls, in other words we can have Common
LISP-like closures.

Out of curiosity, what would be non-Common-Lisp-style closures?

Cheers,
Chris

 

[Antti J Ylikoski]

Out of curiosity, what would be non-Common-Lisp-style closures?

Cheers,
Chris

I understand that a "closure" is something which is typical of
functional programming languages. -- Scheme-style closures, for example.

I don't know Haskell, ML etc. but I do suspect that we could create
closures in those languages as well. Maybe someone more expert than me
can help?

regards, Andy

 

[Antti J Ylikoski]

I understand that a "closure" is something which is typical of
functional programming languages. -- Scheme-style closures, for example.

I don't know Haskell, ML etc. but I do suspect that we could create
closures in those languages as well. Maybe someone more expert than me
can help?

regards, Andy

This is how it is done in standard Common LISP:

-----------------------------------------

;;; Closure with Common LISP.
;;;
;;; Antti J Ylikoski 02-03-2012.

(defun mak-1 ()
(let ((n 0))
#'(lambda () (incf n))))

 

[Arnaud Delobelle]

I understand that a "closure" is something which is typical of functional
programming languages.  -- Scheme-style closures, for example.

I don't know Haskell, ML etc. but I do suspect that we could create closures
in those languages as well.  Maybe someone more expert than me can help?

I think what Chris asking is: what is the feature of Common-Lisp
closures that Python closures share but other languages don't?

I think what he is implying is that there is no such feature. Python
closures are no more "Common-Lisp-style" than they are "Scheme-style"
or "Smalltalk-like" or any other language-like.

 

[Antti J Ylikoski]

I think what Chris asking is: what is the feature of Common-Lisp
closures that Python closures share but other languages don't?

I think what he is implying is that there is no such feature. Python
closures are no more "Common-Lisp-style" than they are "Scheme-style"
or "Smalltalk-like" or any other language-like.

I would say that Python closures are equivalent with Common LISP
closures (except that LAMBDA is more limited in Python, which is a
feature which I don't like.)

Do you maybe mean non-Common-LISP-style closures in Python? I cannot
think of any ones.

kind regards, Andy

 

[Tomasz Rola]

I would say that Python closures are equivalent with Common LISP closures
(except that LAMBDA is more limited in Python, which is a feature which I
don't like.)

Do you maybe mean non-Common-LISP-style closures in Python? I cannot
think of any ones.

kind regards, Andy

AFAIK there is only one style for closure, similar to one style for
square.

There are quite a lot languages implementing closures, and quite a lot try
to imitate them, including C with non-standard extension (without using
those imitations I cannot say if they are good enough).

http://en.wikipedia.org/wiki/Closure_(computer_science)

Wrt lambdas, I really like blocks from Ruby (which AFAIK stem from blocks
in Smalltalk, not sure if they call them "blocks").

http://lesscode.org/2005/07/12/ruby-colored-blocks-in-python/

http://railsguru.org/2010/03/learn-ruby-procs-blocks-lambda/

I mean, myself I am ok with lambdas (using them in languages where lambda
is welcomed and contributing citizen) but blocks in place of lambdas would
be nice to have in Python. Introduction of "with" construct was good IMHO,
but if one likes coding style relying on passing anonymous pieces of code
then Python might not be good choice for this.

On the other hand, one can argue that using anonymous code too much is not
the best style. I am not sure if extensive use of blocks/lambdas really
helps, or if it contributes to "clever" hacks and a source of maintainance
pain. So, perhaps it is good to have it in a few different ways -
like, Ruby, Python and CL - and experiment with them all.

In other words, rather than talking about making Python more like some
other language(s) I think it is much better to learn those other
language(s). If you'd like to try "unlimited" lambda, you might want to
play with Racket, a Scheme superset. Or any other Scheme - it's simple
enough to start coding after a day or two of learning (I mean Fibonaccis
and Erastotenes sieves, not implementing database or web server).

Myself, I would rather have blocks/lambdas and not need them rather than
the other way, but that's just me.

Regards,
Tomasz Rola

--
** A C programmer asked whether computer had Buddha's nature. **
** As the answer, master did "rm -rif" on the programmer's home **
** directory. And then the C programmer became enlightened... **
** **
** Tomasz Rola mailto:[email protected] **

 

[Devin Jeanpierre]

I think what Chris asking is: what is the feature of Common-Lisp
closures that Python closures share but other languages don't?

I think what he is implying is that there is no such feature.  Python
closures are no more "Common-Lisp-style" than they are "Scheme-style"
or "Smalltalk-like" or any other language-like.

"No such feature"? What's that nonlocal thing then? The above function
could not be written that way in Python 2.

Of course maybe we want to put this feature in another category, but
anyway, the function couldn't be written in some languages, even
though they have closures.

-- Devin

 

[John O'Hagan]

On Sat, 04 Feb 2012 02:27:56 +0200

[...]

# Make a Common LISP-like closure with Python.
#
# Antti J Ylikoski 02-03-2012.

def f1():
n = 0
def f2():
nonlocal n
n += 1
return n
return f2

[...]


i. e. we can have several functions with private local states which
are kept between function calls, in other words we can have Common
LISP-like closures.

I'm not sure how naughty this is, but the same thing can be done without using
nonlocal by storing the local state as an attribute of the enclosed function
object:
.... def g():
.... g.count += 1
.... return g.count
.... g.count = 0
.... return g
....3

This way, you can also write to the attribute:
1


John

 

[Antti J Ylikoski]

On Sat, 04 Feb 2012 02:27:56 +0200

[...]

# Make a Common LISP-like closure with Python.
#
# Antti J Ylikoski 02-03-2012.

def f1():
n = 0
def f2():
nonlocal n
n += 1
return n
return f2

[...]


i. e. we can have several functions with private local states which
are kept between function calls, in other words we can have Common
LISP-like closures.

I'm not sure how naughty this is, but the same thing can be done without using
nonlocal by storing the local state as an attribute of the enclosed function
object:
... def g():
... g.count += 1
... return g.count
... g.count = 0
... return g
...3

This way, you can also write to the attribute:
1


John

Yes, I do know that, but then it would not be a closure ))))))))

Andy

 

[Ian Kelly]

Yes, I do know that, but then it would not be a closure ))))))))

Sure it is. Where do you think it looks up the function object?

Cheers,
Ian

 

[Rick Johnson]

In Python textbooks that I have read, it is usually not mentioned that
we can very easily program Common LISP-style closures with Python.  It
is done as follows:
[...]

do my eyes not see nor my ears not hear?
a thread about common Lisp and Xan Lee is not near?
would someone please him wake up and tell him all about,
the thread titled "Common LISP-style closures with Python"
and that he has been left out!

 

[Antti J Ylikoski]

Sure it is. Where do you think it looks up the function object?

Cheers,
Ian

OK, thank you for correcting me.

Andy

 

[John Nagle]

In Python textbooks that I have read, it is usually not mentioned that
we can very easily program Common LISP-style closures with Python. It
is done as follows:

Most dynamic languages have closures. Even Perl and Javascript
have closures. Javascript really needs them, because the "callback"
orientation of Javascript means you often need to package up state
and pass it into a callback. It really has very little to do with
functional programming.

If you want to see a different style of closure, check out Rust,
Mozilla's new language. Rust doesn't have the "spaghetti stack"
needed to implement closures, so it has more limited closure
semantics. It's more like some of the C add-ons for closures,
but sounder.

John Nagle

 

[88888 Dihedral]

在 2012å¹´2月4日星期六UTC+8上åˆ8æ—¶27分56秒，Antti J Ylikoski写é“：

In Python textbooks that I have read, it is usually not mentioned that
we can very easily program Common LISP-style closures with Python. It
is done as follows:

-------------------------------------

# Make a Common LISP-like closure with Python.
#
# Antti J Ylikoski 02-03-2012.

def f1():
n = 0
def f2():
nonlocal n
n += 1
return n
return f2

-------------------------------------

and now we can do:

-------------------------------------


-------------------------------------

i. e. we can have several functions with private local states which
are kept between function calls, in other words we can have Common
LISP-like closures.

yours, Antti J Ylikoski
Helsinki, Finland, the EU

We are not in the 1990's now. A descent CAD or internet application now should be able to support users with at least one or more script
languages easily. Whether it's javascript or java or flash in the browser-based applications, or go, python in the google desktop API,
commercial SW applications to be able to evolve in the long run are not jobs from the publishers and the original writers of the SW packages only.

I don't want to include a big fat compiler in my software,
what else can I do ?

LISP is too fat, too.