Re: Immutability and Python
Am 29.10.2012 16:20 schrieb andrea crotti: Now on one hand I would love to use only immutable data in my code, but on the other hand I wonder if it makes so much sense in Python. You can have both. Many mutable types distinguish between them with their operators. To pick up your example, class NumWrapper(object): def __init__(self, number): self.number = number def __iadd__(self, x): self.number += x return self def __add__(self, x): return NumWrapper(self.number + x) So with number += 1 you keep the same object and modify it, while with number = number + 1 or new_number = number + 1 you create a new object. But more importantly normally classes are way more complicated than my stupid example, so recreating a new object with the modified state might be quite complex. Any comments about this? What do you prefer and why? That's why I generally prefer mutable objects, but it can depend. Thomas -- http://mail.python.org/mailman/listinfo/python-list
Re: Immutability and Python
On Oct 29, 8:20 pm, andrea crotti andrea.crott...@gmail.com wrote: snipped Any comments about this? What do you prefer and why? Im not sure how what the 'prefer' is about -- your specific num wrapper or is it about the general question of choosing mutable or immutable types? If the latter I would suggest you read http://en.wikipedia.org/wiki/Alexander_Stepanov#Criticism_of_OOP [And remember that Stepanov is the author of C++ STL, he is arguably as important in the C++ world as Stroustrup] -- http://mail.python.org/mailman/listinfo/python-list
Re: Immutability and Python
rusi wrote: On Oct 29, 8:20 pm, andrea crotti andrea.crott...@gmail.com wrote: snipped Any comments about this? What do you prefer and why? Im not sure how what the 'prefer' is about -- your specific num wrapper or is it about the general question of choosing mutable or immutable types? If the latter I would suggest you read http://en.wikipedia.org/wiki/Alexander_Stepanov#Criticism_of_OOP [And remember that Stepanov is the author of C++ STL, he is arguably as important in the C++ world as Stroustrup] The usual calls for immutability are not related to OO. They have to do with optimization, and specifically with parallel processing. -- http://mail.python.org/mailman/listinfo/python-list
Re: Immutability and Python
On Oct 31, 1:45 am, Neal Becker ndbeck...@gmail.com wrote: rusi wrote: On Oct 29, 8:20 pm, andrea crotti andrea.crott...@gmail.com wrote: snipped Any comments about this? What do you prefer and why? Im not sure how what the 'prefer' is about -- your specific num wrapper or is it about the general question of choosing mutable or immutable types? If the latter I would suggest you read http://en.wikipedia.org/wiki/Alexander_Stepanov#Criticism_of_OOP [And remember that Stepanov is the author of C++ STL, he is arguably as important in the C++ world as Stroustrup] The usual calls for immutability are not related to OO. They have to do with optimization, and specifically with parallel processing. From the time of Backus' Turing award http://www.thocp.net/biographies/papers/backus_turingaward_lecture.pdf it is standard fare that assignment = imperative programming (which he collectively and polemically called the von Neumann bottleneck) That what he decried as 'conventional programming languages' today applies to OO languages; see http://www.cs.tufts.edu/~nr/backus-lecture.html A more modern viewpoint: -- Object-oriented programming is eliminated entirely from the introductory curriculum, because it is both anti-modular and anti- parallel by its very nature, and hence unsuitable for a modern CS curriculum. A proposed new course on object-oriented design methodology will be offered at the sophomore level for those students who wish to study this topic. from http://existentialtype.wordpress.com/2011/03/15/teaching-fp-to-freshmen/ Call it polemical if you like; noting that that's Carnegie Mellon. -- http://mail.python.org/mailman/listinfo/python-list
Re: Immutability and Python
- Original Message - I have a philosofical doubt about immutability, that arised while doing the SCALA functional programming course. Now suppose I have a simple NumWrapper class, that very stupidly does: class NumWrapper(object): def __init__(self, number): self.number = number and we want to change its state incrementing the number, normally I would do this def increment(self): self.number += 1 But the immutability purists would instead suggest to do this: def increment(self): return NumWrapper(self.number + 1) Now on one hand I would love to use only immutable data in my code, but on the other hand I wonder if it makes so much sense in Python. My impression is that things get more clumsy in the immutable form, for example in the mutable form I would do simply this: number = NumWrapper(1) number.increment() while with immutability I have to do this instead: new_number = number.increment() But more importantly normally classes are way more complicated than my stupid example, so recreating a new object with the modified state might be quite complex. Any comments about this? What do you prefer and why? -- http://mail.python.org/mailman/listinfo/python-list return NumWrapper(self.number + 1) still returns a(nother) mutable object. So what's the point of all this ? JM -- IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you. -- http://mail.python.org/mailman/listinfo/python-list
Re: Immutability and Python
2012/10/29 Jean-Michel Pichavant jeanmic...@sequans.com: return NumWrapper(self.number + 1) still returns a(nother) mutable object. So what's the point of all this ? JM Well sure but it doesn't modify the first object, just creates a new one. There are in general good reasons to do that, for example I can then compose things nicely: num.increment().increment() or I can parallelize operations safely not caring about the order of operations. But while I do this all the time with more functional languages, I don't tend to do exactly the same in Python, because I have the impression that is not worth, but maybe I'm wrong.. -- http://mail.python.org/mailman/listinfo/python-list
Re: Immutability and Python
2012/10/29 andrea crotti andrea.crott...@gmail.com: Well sure but it doesn't modify the first object, just creates a new one. There are in general good reasons to do that, for example I can then compose things nicely: num.increment().increment() or I can parallelize operations safely not caring about the order of operations. But while I do this all the time with more functional languages, I don't tend to do exactly the same in Python, because I have the impression that is not worth, but maybe I'm wrong.. By the way on this topic there is a great talk by the creator of Clojure: http://www.infoq.com/presentations/Value-Values -- http://mail.python.org/mailman/listinfo/python-list
Re: Immutability and Python
On 29/10/2012 15:20, andrea crotti wrote: I have a philosofical doubt about immutability, that arised while doing the SCALA functional programming course. Now suppose I have a simple NumWrapper class, that very stupidly does: class NumWrapper(object): def __init__(self, number): self.number = number and we want to change its state incrementing the number, normally I would do this def increment(self): self.number += 1 But the immutability purists would instead suggest to do this: def increment(self): return NumWrapper(self.number + 1) Now on one hand I would love to use only immutable data in my code, but on the other hand I wonder if it makes so much sense in Python. My impression is that things get more clumsy in the immutable form, for example in the mutable form I would do simply this: number = NumWrapper(1) number.increment() while with immutability I have to do this instead: new_number = number.increment() But more importantly normally classes are way more complicated than my stupid example, so recreating a new object with the modified state might be quite complex. Any comments about this? What do you prefer and why? I prefer practicality beats purity. -- Cheers. Mark Lawrence. -- http://mail.python.org/mailman/listinfo/python-list
Re: Immutability and Python
andrea crotti andrea.crott...@gmail.com writes: and we want to change its state incrementing the number ... the immutability purists would instead suggest to do this: def increment(self): return NumWrapper(self.number + 1) Immutability purists would say that numbers don't have state and if you're trying to change a number's state by incrementing it, that's not immutability. You end up with a rather different programming style than imperative programming, for example using tail recursion (maybe wrapped in an itertools-like higher-order function) instead of indexed loops to iterate over a structure. -- http://mail.python.org/mailman/listinfo/python-list
Re: Immutability and Python
- Original Message - 2012/10/29 Jean-Michel Pichavant jeanmic...@sequans.com: return NumWrapper(self.number + 1) still returns a(nother) mutable object. So what's the point of all this ? JM Well sure but it doesn't modify the first object, just creates a new one. There are in general good reasons to do that, for example I can then compose things nicely: num.increment().increment() or I can parallelize operations safely not caring about the order of operations. But while I do this all the time with more functional languages, I don't tend to do exactly the same in Python, because I have the impression that is not worth, but maybe I'm wrong.. In an OOP language num.increment() is expected to modify the object in place. So I think you're right when you say that functional languages technics do not necessarily apply to Python, because they don't. I would add that what you're trying to suggest in the first post was not really about immutability, immutable objects in python are ... well immutable, they can be used as a dict key for instance, your NumWrapper object cannot. JM -- IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you. -- http://mail.python.org/mailman/listinfo/python-list
Re: Immutability and Python
On Tue, Oct 30, 2012 at 2:55 AM, Paul Rubin no.email@nospam.invalid wrote: andrea crotti andrea.crott...@gmail.com writes: and we want to change its state incrementing the number ... the immutability purists would instead suggest to do this: def increment(self): return NumWrapper(self.number + 1) Immutability purists would say that numbers don't have state and if you're trying to change a number's state by incrementing it, that's not immutability. You end up with a rather different programming style than imperative programming, for example using tail recursion (maybe wrapped in an itertools-like higher-order function) instead of indexed loops to iterate over a structure. In that case, rename increment to next_integer and TYAOOYDAO. [1] You're not changing the state of this number, you're locating the number which has a particular relationship to this one (in the same way that GUI systems generally let you locate the next and previous siblings of any given object). ChrisA [1] there you are, out of your difficulty at once - cf WS Gilbert's Iolanthe -- http://mail.python.org/mailman/listinfo/python-list
Re: Immutability and Python
2012/10/29 Jean-Michel Pichavant jeanmic...@sequans.com: In an OOP language num.increment() is expected to modify the object in place. So I think you're right when you say that functional languages technics do not necessarily apply to Python, because they don't. I would add that what you're trying to suggest in the first post was not really about immutability, immutable objects in python are ... well immutable, they can be used as a dict key for instance, your NumWrapper object cannot. JM Yes right immutable was not the right word, I meant that as a contract with myself I'm never going to modify its state. Also because how doi I make an immutable object in pure Python? But the example with the dictionary is not correct though, because this: In [145]: class C(object): .: def __hash__(self): .: return 42 .: In [146]: d = {C(): 1} works perfectly, but an object of class C can mutate as much as it wants, as my NumWrapper instance.. -- http://mail.python.org/mailman/listinfo/python-list
Re: Immutability and Python
2012/10/29 Chris Angelico ros...@gmail.com: On Tue, Oct 30, 2012 at 2:55 AM, Paul Rubin no.email@nospam.invalid wrote: andrea crotti andrea.crott...@gmail.com writes: and we want to change its state incrementing the number ... the immutability purists would instead suggest to do this: def increment(self): return NumWrapper(self.number + 1) Immutability purists would say that numbers don't have state and if you're trying to change a number's state by incrementing it, that's not immutability. You end up with a rather different programming style than imperative programming, for example using tail recursion (maybe wrapped in an itertools-like higher-order function) instead of indexed loops to iterate over a structure. In that case, rename increment to next_integer and TYAOOYDAO. [1] You're not changing the state of this number, you're locating the number which has a particular relationship to this one (in the same way that GUI systems generally let you locate the next and previous siblings of any given object). ChrisA [1] there you are, out of your difficulty at once - cf WS Gilbert's Iolanthe -- http://mail.python.org/mailman/listinfo/python-list Yes the name should be changed, but the point is that they are both ways to implement the same thing. For example suppose I want to have 10 objects (for some silly reason) that represent the next number, in the first case I would do: numbers = [NumWrapper(orig.number)] * 10 for num in numbers: num.increment() while in the second is as simple as: numbers = [orig.next_number()] * 10 composing things become much easier, but as a downside it's not always so easy and convienient to write code in this way, it probably depends on the use case.. -- http://mail.python.org/mailman/listinfo/python-list
Re: Immutability and Python
andrea crotti andrea.crott...@gmail.com writes: Also because how doi I make an immutable object in pure Python? Numbers in Python are already immutable. What you're really looking for is a programming style where you don't bind any variable more than once. This gives rise to a programming style that Python can support to a certain extent, but for which some other languages are designed from the beginning. You might look at http://learnyouahaskell.com (online book) if you want to try the above approach. -- http://mail.python.org/mailman/listinfo/python-list
Re: Immutability and Python
On 10/29/2012 11:20 AM, andrea crotti wrote: I have a philosofical doubt about immutability, that arised while doing the SCALA functional programming course. In real life, the physical world, things have mutable state, at least down to the atomic level. Do you only want to model mathematical worlds, or also physical worlds. Even mathematically, I do not think there necessarily a problem with mutable collections. The fact that sets are defined by content does not mean that everything has to be. -- Terry Jan Reedy -- http://mail.python.org/mailman/listinfo/python-list
Re: Re: Immutability and Python
On 10/29/2012 12:05 PM, andrea crotti wrote: I meant how do I create new immutables classes myself, I guess that's possible writing C extensions but I don't see in pure Python.. The short answer is: you don't, not really, except by using NamedTuple if that gives you what you want. The longer answer: You can kinda get it somewhat if you define your own __getattribute__/__setattribute__ functions. __setattribute__ of course should never do anything except raise an error (one way or another you'll need to make an exception for your __init__ function of course). __getattribute__ should make sure no mutable references are returned: e.g. you'll probably want to make it so someone can't side-step your setter by saying someobject.__dict__[foo] = bar. (I return a copy of the dict.) It will still be possible to bypass these protections though. To really get true immutability in pure Python, you'll have to inherit from tuple or NamedTuple (which inherits from tuple, I think). You can see some discussion on Stack Overflow and some other places about this; having played around with this a bit, I think it's not worth the hassle and have done the __getattribute__/__setattribute__ thing the couple of times I wanted immutability. Evan signature.asc Description: OpenPGP digital signature -- http://mail.python.org/mailman/listinfo/python-list
Re: Immutability and Python
On 10/29/2012 1:05 PM, andrea crotti wrote: I meant how do I create new immutables classes myself, I guess that's possible writing C extensions but I don't see in pure Python.. If you mean class with immutable instances, mutate new instances in __new__ instead of __init__ and write a custom .__setattr__ that prevents changes thereafter. If you want the class itself to be immutable (after creation), write a custom metaclass. You may also need to think about .__getattribute__, but I never studied the detail completely and have forgotten what I learned. -- Terry Jan Reedy -- http://mail.python.org/mailman/listinfo/python-list
Re: Immutability and Python
On Mon, Oct 29, 2012 at 12:46 PM, Paul Rubin no.email@nospam.invalid wrote: andrea crotti andrea.crott...@gmail.com writes: Also because how doi I make an immutable object in pure Python? Numbers in Python are already immutable. What you're really looking for is a programming style where you don't bind any variable more than once. No, they were looking for a way to create classes whose instances are immutable. Also, immutability has nothing to do with the presence or lack of for loops. -- Devin -- http://mail.python.org/mailman/listinfo/python-list
Re: Immutability and Python
On Mon, Oct 29, 2012 at 10:12 AM, andrea crotti andrea.crott...@gmail.com wrote: Also because how doi I make an immutable object in pure Python? I sometimes use namedtuples for this. from collections import namedtuple MyImmutableClass = namedtuple('MyImmutableClass', 'field1 field2 field3 field4') If you want default arguments then use a factory function. Or if you want the class to have methods, then subclass it: _MyImmutableClass = namedtuple('MyImmutableClass', 'field1 field2 field3 field4') class MyImmutableClass(_MyImmutableClass): def __new__(cls, field1, field2, field3=None, field4=42): return super().__new__(cls, field1, field2, field3, field4) def get_sum(self): return self.field1 + self.field2 Cheers, Ian -- http://mail.python.org/mailman/listinfo/python-list
Re: Immutability and Python
On Tue, Oct 30, 2012 at 6:23 AM, Ian Kelly ian.g.ke...@gmail.com wrote: _MyImmutableClass = namedtuple('MyImmutableClass', 'field1 field2 field3 field4') class MyImmutableClass(_MyImmutableClass): Question: Is it clearer to take advantage of the fact that the base class can be an arbitrary expression? class MyImmutableClass(namedtuple('MyImmutableClass', 'field1 field2 field3 field4')): You lose the unnecessary temporary and triplication of name, but gain instead a rather long line. ChrisA -- http://mail.python.org/mailman/listinfo/python-list
Re: Immutability and Python
On Mon, Oct 29, 2012 at 1:36 PM, Chris Angelico ros...@gmail.com wrote: Question: Is it clearer to take advantage of the fact that the base class can be an arbitrary expression? class MyImmutableClass(namedtuple('MyImmutableClass', 'field1 field2 field3 field4')): You lose the unnecessary temporary and triplication of name, but gain instead a rather long line. I think it's more readable if separated, but YMMV. -- http://mail.python.org/mailman/listinfo/python-list
Re: Immutability and Python
On Mon, 29 Oct 2012 17:05:07 +, andrea crotti wrote: I meant how do I create new immutables classes myself, I guess that's possible writing C extensions but I don't see in pure Python.. Well, you can't *quite* make a truly immutable class in pure-Python, because if *your* Python code can manipulate the class during construction then so can the caller's Python code after construction. The trivial way to make an immutable class in Python is to inherit from an already immutable class and add behaviour but no state: class MyInt(int): def inc(self): return self.__class__(self + 1) Otherwise, you can add private state and rely on the caller not shooting themselves in the foot by accessing single-underscore names, use properties to protect private state, etc. See the source code for collections.namedtuple and decimal.Decimal for some examples. Warning: namedtuple is special, because it uses some runtime exec magic; most immutable classes do not need that. And Decimal is seriously large and complicated. But you can get some ideas from them both. Also, see this: http://northernplanets.blogspot.com.au/2007/01/immutable-instances-in-python.html -- Steven -- http://mail.python.org/mailman/listinfo/python-list
Re: Immutability and Python
On Tue, 30 Oct 2012 06:36:52 +1100, Chris Angelico wrote: On Tue, Oct 30, 2012 at 6:23 AM, Ian Kelly ian.g.ke...@gmail.com wrote: _MyImmutableClass = namedtuple('MyImmutableClass', 'field1 field2 field3 field4') class MyImmutableClass(_MyImmutableClass): Question: Is it clearer to take advantage of the fact that the base class can be an arbitrary expression? class MyImmutableClass(namedtuple('MyImmutableClass', 'field1 field2 field3 field4')): I'm too lazy to google for it, but if you read the examples provided by namedtuple's creator, Raymond Hettinger, that is precisely one of the styles he uses. No need to explicitly declare the base class before using it. -- Steven -- http://mail.python.org/mailman/listinfo/python-list
Re: Immutability and Python
On Mon, Oct 29, 2012 at 3:30 PM, Steven D'Aprano steve+comp.lang.pyt...@pearwood.info wrote: On Mon, 29 Oct 2012 17:05:07 +, andrea crotti wrote: I meant how do I create new immutables classes myself, I guess that's possible writing C extensions but I don't see in pure Python.. Well, you can't *quite* make a truly immutable class in pure-Python, because if *your* Python code can manipulate the class during construction then so can the caller's Python code after construction. The trivial way to make an immutable class in Python is to inherit from an already immutable class and add behaviour but no state: class MyInt(int): def inc(self): return self.__class__(self + 1) Otherwise, you can add private state and rely on the caller not shooting themselves in the foot by accessing single-underscore names, use properties to protect private state, etc. You'd also need to add __slots__ = () to the class definition to make it immutable. Otherwise they still can shoot themselves in the foot by adding new attributes. class MyInt(int): ...def inc(self): ...return self.__class__(self+1) ... a = MyInt() a.b = 1 # Oops. Mutated a. a.b 1 class MyInt(int): ... __slots__ = () ... def inc(self): ... return self.__class__(self + 1) ... a = MyInt() a.b = 1 AttributeError: 'MyInt' object has no attribute 'b' Traceback (most recent call last): File stdin-inspect, line 1, in module AttributeError: 'MyInt' object has no attribute 'b' -- http://mail.python.org/mailman/listinfo/python-list
Re: Immutability and Python
On Mon, 29 Oct 2012 15:20:02 +, andrea crotti wrote: I have a philosofical doubt about immutability, that arised while doing the SCALA functional programming course. Philosophical. Like most words derived from the ancient Greeks, the F sound uses ph rather than f. Now suppose I have a simple NumWrapper class, that very stupidly does: class NumWrapper(object): def __init__(self, number): self.number = number and we want to change its state incrementing the number, normally I would do this def increment(self): self.number += 1 That's a perfectly fine (although incomplete) design for a mutable numeric class. But as the basis of an immutable class, it's lousy. But the immutability purists would instead suggest to do this: def increment(self): return NumWrapper(self.number + 1) Only if they don't know Python very well :-) In this example, the right way to get an immutable class is: class NumWrapper(int): # not exactly a *wrapper* def increment(self): return self.__class__(self + 1) and you're done. Immutability for free, because you don't store state anywhere that pure-Python code can get to it. (Technically, using ctypes you could mutate it, so don't do that.) Here's a sketch of another technique: class MyNum(object): __slots__ = '_num' def __new__(cls, arg): instance = object.__new__(cls) instance._num = int(arg) return instance @property def value(self): return self._num def increment(self): return self.__class__(self.value + 1) Now on one hand I would love to use only immutable data in my code, but on the other hand I wonder if it makes so much sense in Python. You can go a long, long way using only immutable primitives and functional style in Python, and I recommend it. On the other hand, a *purely* functional approach doesn't make a lot of sense for some tasks. Python is not a pure functional language, and doesn't force you to hammer round pegs into the square hole of the functional style. Some problems are best modelled by an object that holds state and can change over time, e.g. a database or a dict. Other problems are best modelled by constants which do not change, but can be replaced by other constants, e.g. numbers. Some problems fall into a grey area, e.g. lists, arrays, sets, sequences, strings. My advice is to always be alert for square pegs in your code, and write them in functional style using immutable instances, but don't be a purist. If you have a round peg, write that part of your code using a mutable instance with in-place mutator methods, and be happy. The beauty of Python is that you can use whichever style suits the problem best. My impression is that things get more clumsy in the immutable form, for example in the mutable form I would do simply this: number = NumWrapper(1) number.increment() while with immutability I have to do this instead: new_number = number.increment() Why is this clumsy? Do you have problems with this? x = 1 y = x+1 -- Steven -- http://mail.python.org/mailman/listinfo/python-list
Re: Immutability and Python
On Mon, 29 Oct 2012 15:45:59 -0700, Chris Kaynor wrote: On Mon, Oct 29, 2012 at 3:30 PM, Steven D'Aprano steve+comp.lang.pyt...@pearwood.info wrote: On Mon, 29 Oct 2012 17:05:07 +, andrea crotti wrote: I meant how do I create new immutables classes myself, I guess that's possible writing C extensions but I don't see in pure Python.. Well, you can't *quite* make a truly immutable class in pure-Python, because if *your* Python code can manipulate the class during construction then so can the caller's Python code after construction. The trivial way to make an immutable class in Python is to inherit from an already immutable class and add behaviour but no state: class MyInt(int): def inc(self): return self.__class__(self + 1) Otherwise, you can add private state and rely on the caller not shooting themselves in the foot by accessing single-underscore names, use properties to protect private state, etc. You'd also need to add __slots__ = () to the class definition to make it immutable. Otherwise they still can shoot themselves in the foot by adding new attributes. Doctor, it hurts when I do this. Then don't do that. I'm not a big fan of preventatively using __slots__ merely to prevent the caller from tagging an object with extra data. Why do you care if the caller sticks a postit note on the object? It doesn't hurt the object, and if the caller loses track of which object has a postit note, that's their responsibility, not yours. I often wish I could sick an attribute on built-ins, e.g. after calculating some numeric result as a float, stick an error estimate on it. Callers who care about the error estimate can inspect it; those who don't, will never even notice it. If you have a good reason for using __slots__, then go right ahead. Otherwise, don't be paternalistic. This is Python, we have the right to shoot ourselves in the foot if we like. -- Steven -- http://mail.python.org/mailman/listinfo/python-list