static variables in Python?

[kj]

Yet another noob question...

Is there a way to mimic C's static variables in Python? Or something
like it? The idea is to equip a given function with a set of
constants that belong only to it, so as not to clutter the global
namespace with variables that are not needed elsewhere.

For example, in Perl one can define a function foo like this

*foo = do {
my $x = expensive_call();
sub {
return do_stuff_with( $x, @_ );
}
};

In this case, foo is defined by assigning to it a closure that has
an associated variable, $x, in its scope.

Is there an equivalent in Python?

Thanks!

kynn

 

[kj]

First names in Python are just that, names that point to objects. Those objects
can contain any type of information including other objects. They are NOT
buckets where things are stored.

1) Names (variables in Perl/C) defined within a Python function are placed in
its local namespace. They are not visible in the global namespace.

2) Yes you can have a local name point to a global. This is often used in
classes with attributes because looking up local is somewhat quicker than
looking up the class attribute.

def foo():
x = expensive_call
return do_stuff_with(x())

Maybe I'm missing your point, the goal is to have a "runtime
constant" associated with the function. In the your definition of
foo, expensive_call gets called every time that foo gets called;
this is what I'm trying to avoid!

Maybe it's easier to see what I mean with JavaScript:

function foo() {
if (foo.x === undefined) foo.x = expensive_call();
return do_stuff_with(foo.x);
}

Here, expensive_call is called only once (assuming it never returns
undefined).

OK, I guess that in Python the only way to do what I want to do is
with objects...

kynn

 

[Colin J. Williams]

Maybe I'm missing your point, the goal is to have a "runtime
constant" associated with the function. In the your definition of
foo, expensive_call gets called every time that foo gets called;
this is what I'm trying to avoid!

Maybe it's easier to see what I mean with JavaScript:

function foo() {
if (foo.x === undefined) foo.x = expensive_call();
return do_stuff_with(foo.x);
}

Here, expensive_call is called only once (assuming it never returns
undefined).

OK, I guess that in Python the only way to do what I want to do is
with objects...

kynn

You might consider using a singleton class.

Colin W.

 

[bearophileHUGS]

kj:

OK, I guess that in Python the only way to do what I want to do
is with objects...

There are other ways, like assigning the value out of the function,
because Python functions too are objects:

def iamslow():
return 100
def foo(x):
return x + foo.y
foo.y = iamslow() # slow computation
print foo(1)
print foo(2)

Output is:
101
102

Another way is this, a bit more clean, with the same output:

def iamslow():
return 100
def foo(x, y=iamslow()):
return x + y
print foo(1)
print foo(2)

But I suggest you to use a class in this situation, it's often the way
that will keep your code more bug-free, and more readable by near-
casual readers too. Python philosophy asks you to write readable code
instead of clever code when possible, this is a difference from Perl,
I presume.

Bye,
bearophile

 

[Bruce Frederiksen]

[snip]

Maybe it's easier to see what I mean with JavaScript:

function foo() {
if (foo.x === undefined) foo.x = expensive_call();
return do_stuff_with(foo.x);
}

def foo():
if not hasattr(foo, 'x'): foo.x = expensive_call()
return do_stuff_with(foo.x)

or, maybe just define foo in two steps:

def foo():
return do_stuff_with(foo.x)

foo.x = expensive_call()

 

[Russ P.]

Yet another noob question...

Is there a way to mimic C's static variables in Python? Or something
like it? The idea is to equip a given function with a set of
constants that belong only to it, so as not to clutter the global
namespace with variables that are not needed elsewhere.

For example, in Perl one can define a function foo like this

*foo = do {
my $x = expensive_call();
sub {
return do_stuff_with( $x, @_ );
}

};

In this case, foo is defined by assigning to it a closure that has
an associated variable, $x, in its scope.

Is there an equivalent in Python?

Thanks!

kynn

If the constant parameters are really only needed in one particular
function, you can use default function arguments. An added benefit is
that you can override them with another value if necessary.

def fun(x, y, parameter1=0, parameter2=1):
...

 

[Russ P.]

If the constant parameters are really only needed in one particular
function, you can use default function arguments. An added benefit is
that you can override them with another value if necessary.

def fun(x, y, parameter1=0, parameter2=1):
...

I should add that the parameters need not be literal numbers. They can
be computed values as well. They will be computed only once, on the
first pass through the function definition, which I presume is exactly
what you want.

I think this is the simplest solution to the problem you posed.

 

[John Machin]

I should add that the parameters need not be literal numbers. They can
be computed values as well. They will be computed only once, on the
first pass through the function definition, which I presume is exactly
what you want.

I think this is the simplest solution to the problem you posed.

Here's a real-life example, where the second and third args are run-
time constants:

def unescape(s,
subber=re.compile(r'_x[0-9A-Fa-f]{4,4}_').sub,
repl=lambda mobj: unichr(int(mobj.group(0)[2:6], 16)),
):
if "_" in s:
return subber(repl, s)
return s
# The if test is just an optimisation that unfortunately the re module
doesn't nut out for itself.

Cheers,
John

 

[castironpi]

kj:

There are other ways, like assigning the value out of the function,
because Python functions too are objects:
...
But I suggest you to use a class in this situation, it's often the way
that will keep your code more bug-free, and more readable by near-
casual readers too. Python philosophy asks you to write readable code
instead of clever code when possible, this is a difference from Perl,
I presume.

Bye,
bearophile

Here's a solution using decorators, I like it, but I'm biased:

def staticAttrs(**kwds):
        """
        Adds attributes to a function, akin to c-style
        "static" variables
        """

        def _decorator(fcn):
                for k in kwds:
                        setattr(fcn, k, kwds[k])
                return fcn
        return _decorator

@staticAttrs(n=0)
def rememberCalls():
        """
        >>> rememberCalls()
        0
        >>> rememberCalls()
        1
        >>> rememberCalls()
        2
        """
        print rememberCalls.n
        rememberCalls.n += 1

~Scott

I like it too. It also thought of (implementation not shown):

@has_locals
def rememberCalls( self ):
self.val= 0
self.ref= object( )

where self is preserved between calls and is an instance of a custom
class, possibly empty. If you want more than one, but still
preserved:

rememberCallsA= has_locals( rememberCalls )
rememberCallsB= has_locals( rememberCalls )

You might want to make self a small and lightweight dict-only object:

@has_locals
def rememberCalls( dic ):
dic['val']= 0
dic['ref']= object( )

 

[Daniel da Silva]

This is the solution I suggest. It is fairly trivial, and works by
introducing the "self.static" namespace for a class's static
variables, in contrast to "self" for the class's instance variables.

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

class Static(object): pass
personStatic = Static()

class Person:
static = personStatic

def __init__(self, name, age):
self.name = name
self.age = age

def setVersion(self, version):
self.static.version = version

def getVersion(self):
return self.static.version
-----------------------------------

Daniel

 

[Alan Franzoni]

kj was kind enough to say:

In this case, foo is defined by assigning to it a closure that has
an associated variable, $x, in its scope.

Is there an equivalent in Python?

There've been plenty of answers, and I'm not absolutely sure about what you
want... but closures are available in Python as well and you can use them,
and by combining them through the partial module you can get a sort of
closure factory:

from functools import partial

def getfunc(expensive_call, myfunc):
val = expensive_call()

f = partial(myfunc, val)

return f



you can then do something like that:
6





--
Alan Franzoni <[email protected]>
-
Remove .xyz from my email in order to contact me.
-
GPG Key Fingerprint:
5C77 9DC3 BD5B 3A28 E7BC 921A 0255 42AA FE06 8F3E

 

[Rhamphoryncus]

Yet another noob question...

Is there a way to mimic C's static variables in Python?  Or something
like it?  The idea is to equip a given function with a set of
constants that belong only to it, so as not to clutter the global
namespace with variables that are not needed elsewhere.

I'd go ahead and use globals. If these really are constant you should
just name them clearly (and possibly use all caps).

If they have a possibility of becoming non-constant in the future,
write a class. No fancy tricks needed to store state.

 

[5lvqbwl02]

Yet another noob question...

Is there a way to mimic C's static variables in Python? Or something
like it? The idea is to equip a given function with a set of
constants that belong only to it, so as not to clutter the global
namespace with variables that are not needed elsewhere.

For example, in Perl one can define a function foo like this

*foo = do {
my $x = expensive_call();
sub {
return do_stuff_with( $x, @_ );
}
};

In this case, foo is defined by assigning to it a closure that has
an associated variable, $x, in its scope.

Is there an equivalent in Python?

I found the following link addressing this problem several different
ways. My favorite (trickiest) way is using decorators...

http://www.daniweb.com/code/snippet501.html

 

[Stephan Schulz]

Yet another noob question...

Is there a way to mimic C's static variables in Python? Or something
like it? The idea is to equip a given function with a set of
constants that belong only to it, so as not to clutter the global
namespace with variables that are not needed elsewhere.

I know I'm coming late to the discussion, but the most natural way for
me would be to simulate the function via a callable object:


class hidden(object):
def __init__(self):
self.static_var = 0

def __call__(self):
self.static_var+=1
return self.static_var

fun_with_state = hidden()

4

Bye,

Stephan