Re: Dynamically replacing an objects __class__; is it safe?

[jladasky]

On Thursday, March 16, 2017 at 7:07:17 PM UTC-7, Deborah Swanson wrote:
> Steve D'Aprano wrote,on March 16, 2017 5:07 AM
> > 
> > On Thu, 16 Mar 2017 09:03 am, Gregory Ewing wrote:
> > 
> > > Steve D'Aprano wrote:
> > >> You probably can't make a whale fly just by changing the class to 
> > >> bird. It will need wings, and feathers, at the very least.
> > > 
> > > Some things succeed in flying with neither wings nor feathers. 
> > > Helicopters, for example.
> > 
> > For some definition of "no wings".
> > 
> > But regardless of rockets, balloons, helicopters, Boeing 747s 
> > and Dr Strange's mystical Cloak Of Levitation, *birds* 
> > require wings and feathers to fly. If you just paint "BIRD" 
> > on the side of a whale, it won't get off the ground, and if 
> > you do manage to get it airborne (via a catapult,
> > perhaps) it will just come down with a rather large splat.
> > 
> > The point is that dynamically swapping the class of an 
> > existing instance at runtime is *not* just a way of doing 
> > duck-typing. It really does matter if your duck flies by 
> > flapping feathered wings or by blasting an exhaust of hot 
> > gasses out of its rear end at high speed.
> > 
> > With duck-typing, you don't care whether you have a duck or a 
> > goose, so long as they both can fly: you don't care *how* it 
> > flies, so long as it does, and even a rocket-propelled 
> > balloon will be fine.
> > 
> > But when you dynamically swap out __class__ (I think 
> > Objective-C "swizzling" is conceptually equivalent, so I'll 
> > call it by that word) you have to care about the 
> > implementation. The whole point of swizzling is that it 
> > allows you to swap out one implementation ("run forward and 
> > flap your wings") with another implementation ("light the 
> > blue touch paper"). But if you do that, you have to care 
> > about implementation details. There's no point in setting 
> > your swizzled fly() method to that of Rocket if your instance 
> > doesn't have blue touch paper to light.
> > 
> > With duck-typing, making the methods work isn't your 
> > responsibility. But when you swizzle, you are responsible for 
> > making sure that the instance provides whatever the methods 
> > need to work.
> > 
> > A very old but good example of Python swizzling is here:
> > http://code.activestate.com/recipes/68429-ring-buffer/
> 
> -- 
> Steve
> "Cheer up," they said, "things could be worse." So I cheered up, and
> sure enough, things got worse.
> 
> Condolences, Steve.  Nobody on this thread wants to plumb the depths of
> Python swizzling. Or anything else Python, so it appears.

Hey Deborah,

If you're getting discouraged by a certain spammer who posts to this newsgroup 
in ALL CAPS and ITALIAN, don't let him beat you.  Go find newsreader software 
that allows you to define filters and killfiles.  It will help.

If you read further into that article about the RingBuffer class and 
"swizzling", Steve Alexander suggests a method that I have used myself, and 
which I consider to be generally superior: conditionally rebinding the names of 
methods you want to change within a class, rather than overriding __class__ 
itself.

I suppose that each approach has its merits.  If you want the object's type to 
advertise that a condition has changed, rebinding __class__ is good.  If you 
have many methods to change, and you don't like long lists of similarly-named 
functions in one class, rebinding __class__ could also be good.

I don't encounter those conditions much myself.  In the RingBufferFull case, 
only two methods need to change, .append() and .get().  Both methods are short. 
 Both methods' eventual replacements, which we might name ._full_append() and 
._get(), are also short.  I would do it Steve Alexander's way.
-- 
https://mail.python.org/mailman/listinfo/python-list

Re: Dynamically replacing an objects __class__; is it safe?

[Chris Angelico]

On Sat, Mar 18, 2017 at 8:24 AM,   wrote:
> On Thursday, March 16, 2017 at 9:27:56 PM UTC-7, Gregory Ewing wrote:
>> Chris Angelico wrote:
>> > Maybe what the ISS does isn't flying - it's falling with style?
>>
>> Yep. They didn't really launch it into orbit with rockets,
>> that was all faked. They actually hauled it up there with
>> a crane, let it go and pulled the earth away at the last
>> moment.
>>
>> --
>> Greg
>
> Chris, this is evidence you need to stick to Flying Circus references when 
> attempting to construct a joke in this newsgroup.  Buzz Lightyear won't cut 
> it.

And cue the discussion of whether Joke.__init__() is the constructor or not.

ChrisA
-- 
https://mail.python.org/mailman/listinfo/python-list

Re: Dynamically replacing an objects __class__; is it safe?

[jladasky]

On Thursday, March 16, 2017 at 9:27:56 PM UTC-7, Gregory Ewing wrote:
> Chris Angelico wrote:
> > Maybe what the ISS does isn't flying - it's falling with style?
> 
> Yep. They didn't really launch it into orbit with rockets,
> that was all faked. They actually hauled it up there with
> a crane, let it go and pulled the earth away at the last
> moment.
> 
> -- 
> Greg

Chris, this is evidence you need to stick to Flying Circus references when 
attempting to construct a joke in this newsgroup.  Buzz Lightyear won't cut it.
-- 
https://mail.python.org/mailman/listinfo/python-list

Re: Dynamically replacing an objects __class__; is it safe?

[Tony van der Hoff]

On 17/03/17 04:18, Gregory Ewing wrote:

Dennis Lee Bieber wrote:

I'd say satellites do "not" fly, as they have no force/action
opposing
the fall caused by the pull of gravity.


Arrows, bullets, thrown stones, etc. are often said to be
flying.

Seems to me the word gets applied to anything that is
moving while not contacting the ground.


Yep, like a Ferrari on a motorway


--
Tony van der Hoff| mailto:t...@vanderhoff.org
Buckinghamshire, England |
--
https://mail.python.org/mailman/listinfo/python-list

Re: Dynamically replacing an objects __class__; is it safe?

[Gregory Ewing]

Chris Angelico wrote:

Maybe what the ISS does isn't flying - it's falling with style?


Yep. They didn't really launch it into orbit with rockets,
that was all faked. They actually hauled it up there with
a crane, let it go and pulled the earth away at the last
moment.

--
Greg
--
https://mail.python.org/mailman/listinfo/python-list

Re: Dynamically replacing an objects __class__; is it safe?

[Gregory Ewing]

Dennis Lee Bieber wrote:

I'd say satellites do "not" fly, as they have no force/action opposing
the fall caused by the pull of gravity.


Arrows, bullets, thrown stones, etc. are often said to be
flying.

Seems to me the word gets applied to anything that is
moving while not contacting the ground.

--
Greg
--
https://mail.python.org/mailman/listinfo/python-list

Re: Dynamically replacing an objects __class__; is it safe?

[Chris Angelico]

On Fri, Mar 17, 2017 at 1:52 PM, Ian Kelly  wrote:
> On Thu, Mar 16, 2017 at 7:56 PM, Dennis Lee Bieber
>  wrote:
>> On Fri, 17 Mar 2017 10:03:04 +1300, Gregory Ewing
>>  declaimed the following:
>>
>>>Chris Angelico wrote:
 I just asked my father, and he believes that a satellite DOES fly, on
 the basis that it is perpetually falling and forever missing the
 ground.
>>>
>>>Also, NASA people talk about "flying" the ISS, so it seems your
>>>father is in good company.
>>
>> Most likely they are referring to using gyroscopes (reaction wheels),
>> and maneuvering thrusters to maintain orientation relative to the sun, so
>> the solar panels stay fully lit.
>
> The ISS orbit is low enough that they also have to do regular
> (approximately monthly) boosts to counteract the gradual decay.

Maybe what the ISS does isn't flying - it's falling with style?

ChrisA
-- 
https://mail.python.org/mailman/listinfo/python-list

Re: Dynamically replacing an objects __class__; is it safe?

[Ian Kelly]

On Thu, Mar 16, 2017 at 7:56 PM, Dennis Lee Bieber
 wrote:
> On Fri, 17 Mar 2017 10:03:04 +1300, Gregory Ewing
>  declaimed the following:
>
>>Chris Angelico wrote:
>>> I just asked my father, and he believes that a satellite DOES fly, on
>>> the basis that it is perpetually falling and forever missing the
>>> ground.
>>
>>Also, NASA people talk about "flying" the ISS, so it seems your
>>father is in good company.
>
> Most likely they are referring to using gyroscopes (reaction wheels),
> and maneuvering thrusters to maintain orientation relative to the sun, so
> the solar panels stay fully lit.

The ISS orbit is low enough that they also have to do regular
(approximately monthly) boosts to counteract the gradual decay.
-- 
https://mail.python.org/mailman/listinfo/python-list

RE: Dynamically replacing an objects __class__; is it safe?

[Deborah Swanson]

Steve D'Aprano wrote,on March 16, 2017 5:07 AM
> 
> On Thu, 16 Mar 2017 09:03 am, Gregory Ewing wrote:
> 
> > Steve D'Aprano wrote:
> >> You probably can't make a whale fly just by changing the class to 
> >> bird. It will need wings, and feathers, at the very least.
> > 
> > Some things succeed in flying with neither wings nor feathers. 
> > Helicopters, for example.
> 
> For some definition of "no wings".
> 
> But regardless of rockets, balloons, helicopters, Boeing 747s 
> and Dr Strange's mystical Cloak Of Levitation, *birds* 
> require wings and feathers to fly. If you just paint "BIRD" 
> on the side of a whale, it won't get off the ground, and if 
> you do manage to get it airborne (via a catapult,
> perhaps) it will just come down with a rather large splat.
> 
> The point is that dynamically swapping the class of an 
> existing instance at runtime is *not* just a way of doing 
> duck-typing. It really does matter if your duck flies by 
> flapping feathered wings or by blasting an exhaust of hot 
> gasses out of its rear end at high speed.
> 
> With duck-typing, you don't care whether you have a duck or a 
> goose, so long as they both can fly: you don't care *how* it 
> flies, so long as it does, and even a rocket-propelled 
> balloon will be fine.
> 
> But when you dynamically swap out __class__ (I think 
> Objective-C "swizzling" is conceptually equivalent, so I'll 
> call it by that word) you have to care about the 
> implementation. The whole point of swizzling is that it 
> allows you to swap out one implementation ("run forward and 
> flap your wings") with another implementation ("light the 
> blue touch paper"). But if you do that, you have to care 
> about implementation details. There's no point in setting 
> your swizzled fly() method to that of Rocket if your instance 
> doesn't have blue touch paper to light.
> 
> With duck-typing, making the methods work isn't your 
> responsibility. But when you swizzle, you are responsible for 
> making sure that the instance provides whatever the methods 
> need to work.
> 
> A very old but good example of Python swizzling is here:
> http://code.activestate.com/recipes/68429-ring-buffer/

-- 
Steve
"Cheer up," they said, "things could be worse." So I cheered up, and
sure enough, things got worse.

Condolences, Steve.  Nobody on this thread wants to plumb the depths of
Python swizzling. Or anything else Python, so it appears.


-- 
https://mail.python.org/mailman/listinfo/python-list

Re: Dynamically replacing an objects __class__; is it safe?

[Gregory Ewing]

Chris Angelico wrote:

I just asked my father, and he believes that a satellite DOES fly, on
the basis that it is perpetually falling and forever missing the
ground.


Also, NASA people talk about "flying" the ISS, so it seems your
father is in good company.

--
Greg
--
https://mail.python.org/mailman/listinfo/python-list

Re: Dynamically replacing an objects __class__; is it safe?

[alister]

On Thu, 16 Mar 2017 14:04:19 +, Jon Ribbens wrote:

> On 2017-03-16, Robin Becker  wrote:
>> On 15/03/2017 13:53, Steve D'Aprano wrote:
>>> You probably can't make a whale fly just by changing the class to
>>> bird. It will need wings, and feathers, at the very least.
>>
>> the whale in the Hitchhiker's Guide found itself flying without
>> feathers or wings
> 
> Well, only if you redefine "flying" to include "accelerating to terminal
> velocity straight downwards under the force of gravity"...

Arthur himself was a regular flyer.




-- 
genius, n.:
Person clever enough to be born in the right place at the right
time of the right sex and to follow up this advantage by saying
all the right things to all the right people.
-- 
https://mail.python.org/mailman/listinfo/python-list

Re: Dynamically replacing an objects __class__; is it safe?

[Chris Green]

Robin Becker  wrote:
> On 16/03/2017 14:03, D'Arcy Cain wrote:
> > On 2017-03-16 09:45 AM, Robin Becker wrote:
> >> On 15/03/2017 13:53, Steve D'Aprano wrote:
> >>> You probably can't make a whale fly just by changing the class to
> >>> bird. It
> >>> will need wings, and feathers, at the very least.
> >>
> >> the whale in the Hitchhiker's Guide found itself flying without feathers
> >> or wings
> >
> > Falling is not the same as flying unless you accidentally miss the ground.
> >
> well in English an arrow flies as well as time, the whale was very interested 
> in 
> the approaching ground and perhaps forgot to conjure up virtual wings or 
> rockets. In any case satellites miss the ground, but even their motion cannot 
> be 
> classed as flying with wings so I think this discussion has become irrelevant.

Time flies like an arrow
Fruit flies like a banana

-- 
Chris Green
·
-- 
https://mail.python.org/mailman/listinfo/python-list

Re: Dynamically replacing an objects __class__; is it safe?

[Chris Angelico]

On Fri, Mar 17, 2017 at 1:21 AM, Robin Becker  wrote:
>> Falling is not the same as flying unless you accidentally miss the ground.
>>
> well in English an arrow flies as well as time, the whale was very
> interested in the approaching ground and perhaps forgot to conjure up
> virtual wings or rockets. In any case satellites miss the ground, but even
> their motion cannot be classed as flying with wings so I think this
> discussion has become irrelevant.

I just asked my father, and he believes that a satellite DOES fly, on
the basis that it is perpetually falling and forever missing the
ground.

ChrisA
-- 
https://mail.python.org/mailman/listinfo/python-list

Re: Dynamically replacing an objects __class__; is it safe?

[Robin Becker]

On 16/03/2017 14:03, D'Arcy Cain wrote:

On 2017-03-16 09:45 AM, Robin Becker wrote:

On 15/03/2017 13:53, Steve D'Aprano wrote:

You probably can't make a whale fly just by changing the class to
bird. It
will need wings, and feathers, at the very least.


the whale in the Hitchhiker's Guide found itself flying without feathers
or wings


Falling is not the same as flying unless you accidentally miss the ground.

well in English an arrow flies as well as time, the whale was very interested in 
the approaching ground and perhaps forgot to conjure up virtual wings or 
rockets. In any case satellites miss the ground, but even their motion cannot be 
classed as flying with wings so I think this discussion has become irrelevant.

--
Robin Becker

--
https://mail.python.org/mailman/listinfo/python-list

Re: Dynamically replacing an objects __class__; is it safe?

[Tony van der Hoff]

On 16/03/17 13:45, Robin Becker wrote:

On 15/03/2017 13:53, Steve D'Aprano wrote:

You probably can't make a whale fly just by changing the class to
bird. It
will need wings, and feathers, at the very least.


the whale in the Hitchhiker's Guide found itself flying without feathers
or wings


It didn't so much fly, as plummet.

--
Tony van der Hoff| mailto:t...@vanderhoff.org
Buckinghamshire, England |
--
https://mail.python.org/mailman/listinfo/python-list

Re: Dynamically replacing an objects __class__; is it safe?

[Jon Ribbens]

On 2017-03-16, Robin Becker  wrote:
> On 15/03/2017 13:53, Steve D'Aprano wrote:
>> You probably can't make a whale fly just by changing the class to bird. It
>> will need wings, and feathers, at the very least.
>
> the whale in the Hitchhiker's Guide found itself flying without
> feathers or wings

Well, only if you redefine "flying" to include "accelerating to
terminal velocity straight downwards under the force of gravity"...
-- 
https://mail.python.org/mailman/listinfo/python-list

Re: Dynamically replacing an objects __class__; is it safe?

[D'Arcy Cain]

On 2017-03-16 09:45 AM, Robin Becker wrote:

On 15/03/2017 13:53, Steve D'Aprano wrote:

You probably can't make a whale fly just by changing the class to
bird. It
will need wings, and feathers, at the very least.


the whale in the Hitchhiker's Guide found itself flying without feathers
or wings


Falling is not the same as flying unless you accidentally miss the ground.

--
D'Arcy J.M. Cain
Vybe Networks Inc.
http://www.VybeNetworks.com/
IM:da...@vex.net VoIP: sip:da...@vybenetworks.com
--
https://mail.python.org/mailman/listinfo/python-list

Re: Dynamically replacing an objects __class__; is it safe?

[Robin Becker]

On 15/03/2017 13:53, Steve D'Aprano wrote:

You probably can't make a whale fly just by changing the class to bird. It
will need wings, and feathers, at the very least.


the whale in the Hitchhiker's Guide found itself flying without feathers or 
wings

--
Robin Becker

--
https://mail.python.org/mailman/listinfo/python-list

Re: Dynamically replacing an objects __class__; is it safe?

[Steve D'Aprano]

On Thu, 16 Mar 2017 09:03 am, Gregory Ewing wrote:

> Steve D'Aprano wrote:
>> You probably can't make a whale fly just by changing the class to bird.
>> It will need wings, and feathers, at the very least.
> 
> Some things succeed in flying with neither wings nor feathers.
> Helicopters, for example.

For some definition of "no wings".

But regardless of rockets, balloons, helicopters, Boeing 747s and Dr
Strange's mystical Cloak Of Levitation, *birds* require wings and feathers
to fly. If you just paint "BIRD" on the side of a whale, it won't get off
the ground, and if you do manage to get it airborne (via a catapult,
perhaps) it will just come down with a rather large splat.

The point is that dynamically swapping the class of an existing instance at
runtime is *not* just a way of doing duck-typing. It really does matter if
your duck flies by flapping feathered wings or by blasting an exhaust of
hot gasses out of its rear end at high speed.

With duck-typing, you don't care whether you have a duck or a goose, so long
as they both can fly: you don't care *how* it flies, so long as it does,
and even a rocket-propelled balloon will be fine.

But when you dynamically swap out __class__ (I think Objective-C "swizzling"
is conceptually equivalent, so I'll call it by that word) you have to care
about the implementation. The whole point of swizzling is that it allows
you to swap out one implementation ("run forward and flap your wings") with
another implementation ("light the blue touch paper"). But if you do that,
you have to care about implementation details. There's no point in setting
your swizzled fly() method to that of Rocket if your instance doesn't have
blue touch paper to light.

With duck-typing, making the methods work isn't your responsibility. But
when you swizzle, you are responsible for making sure that the instance
provides whatever the methods need to work.

A very old but good example of Python swizzling is here:

http://code.activestate.com/recipes/68429-ring-buffer/



-- 
Steve
“Cheer up,” they said, “things could be worse.” So I cheered up, and sure
enough, things got worse.

-- 
https://mail.python.org/mailman/listinfo/python-list

Re: Dynamically replacing an objects __class__; is it safe?

[alister]

On Wed, 15 Mar 2017 19:10:17 -0700, Deborah Swanson wrote:

> MRAB wrote, on Wednesday, March 15, 2017 3:19 PM
>> 
>> On 2017-03-15 22:03, Gregory Ewing wrote:
>> > Steve D'Aprano wrote:
>> >> You probably can't make a whale fly just by changing the class to
>> >> bird. It will need wings, and feathers, at the very least.
>> >
>> > Some things succeed in flying with neither wings nor feathers.
>> > Helicopters, for example.
>> >
>> Could you argue that the blades were a kind of wing? After all, they
>> rely on the same principle of moving through the air to produce lift.
>> 
>> Balloons, on the other hand, ... :-)
> 
> or kites...  ;)

no kites are a wing, although they are dependant on the string to hold 
them up.




-- 
Too many people are thinking of security instead of opportunity.  They 
seem
more afraid of life than death.
-- James F. Byrnes
-- 
https://mail.python.org/mailman/listinfo/python-list

Re: Dynamically replacing an objects __class__; is it safe?

[alister]

On Thu, 16 Mar 2017 11:03:56 +1300, Gregory Ewing wrote:

> Steve D'Aprano wrote:
>> You probably can't make a whale fly just by changing the class to bird.
>> It will need wings, and feathers, at the very least.
> 
> Some things succeed in flying with neither wings nor feathers.
> Helicopters, for example.

helicopters DO have wings.

that is why they are technical referred to as ROTARY WING Aircraft ;-)



-- 
Boy, that crayon sure did hurt!
-- 
https://mail.python.org/mailman/listinfo/python-list

Re: Dynamically replacing an objects __class__; is it safe?

[Gregory Ewing]

MRAB wrote:
Could you argue that the blades were a kind of wing? After all, they 
rely on the same principle of moving through the air to produce lift.


Balloons, on the other hand, ... :-)


Also rockets.

--
Greg
--
https://mail.python.org/mailman/listinfo/python-list

RE: Dynamically replacing an objects __class__; is it safe?

[Deborah Swanson]

MRAB wrote, on Wednesday, March 15, 2017 3:19 PM
> 
> On 2017-03-15 22:03, Gregory Ewing wrote:
> > Steve D'Aprano wrote:
> >> You probably can't make a whale fly just by changing the class to 
> >> bird. It will need wings, and feathers, at the very least.
> >
> > Some things succeed in flying with neither wings nor feathers. 
> > Helicopters, for example.
> >
> Could you argue that the blades were a kind of wing? After all, they 
> rely on the same principle of moving through the air to produce lift.
> 
> Balloons, on the other hand, ... :-)

or kites...  ;)

-- 
https://mail.python.org/mailman/listinfo/python-list

Re: Dynamically replacing an objects __class__; is it safe?

[Chris Angelico]

On Thu, Mar 16, 2017 at 9:19 AM, MRAB  wrote:
> On 2017-03-15 22:03, Gregory Ewing wrote:
>>
>> Steve D'Aprano wrote:
>>>
>>> You probably can't make a whale fly just by changing the class to bird.
>>> It
>>> will need wings, and feathers, at the very least.
>>
>>
>> Some things succeed in flying with neither wings nor feathers.
>> Helicopters, for example.
>>
> Could you argue that the blades were a kind of wing? After all, they rely on
> the same principle of moving through the air to produce lift.
>
> Balloons, on the other hand, ... :-)

Certainly not. If they were, you'd move (or spin) them faster to go
up, and slower to go down. But you don't. So it's all about the
ugliness. You twist the blades up a bit to make the helicopter uglier.

ChrisA
-- 
https://mail.python.org/mailman/listinfo/python-list

Re: Dynamically replacing an objects __class__; is it safe?

[MRAB]

On 2017-03-15 22:03, Gregory Ewing wrote:

Steve D'Aprano wrote:

You probably can't make a whale fly just by changing the class to bird. It
will need wings, and feathers, at the very least.


Some things succeed in flying with neither wings nor feathers.
Helicopters, for example.

Could you argue that the blades were a kind of wing? After all, they 
rely on the same principle of moving through the air to produce lift.


Balloons, on the other hand, ... :-)

--
https://mail.python.org/mailman/listinfo/python-list

Re: Dynamically replacing an objects __class__; is it safe?

[Chris Angelico]

On Thu, Mar 16, 2017 at 9:03 AM, Gregory Ewing
 wrote:
> Steve D'Aprano wrote:
>>
>> You probably can't make a whale fly just by changing the class to bird. It
>> will need wings, and feathers, at the very least.
>
>
> Some things succeed in flying with neither wings nor feathers.
> Helicopters, for example.

Yeah but they do it by being so ugly that the earth repels them. I'm
not sure that whale.ugliness is valid or high enough.

ChrisA
-- 
https://mail.python.org/mailman/listinfo/python-list

Re: Dynamically replacing an objects __class__; is it safe?

[Gregory Ewing]

Steve D'Aprano wrote:

You probably can't make a whale fly just by changing the class to bird. It
will need wings, and feathers, at the very least.


Some things succeed in flying with neither wings nor feathers.
Helicopters, for example.

--
Greg
--
https://mail.python.org/mailman/listinfo/python-list

Re: Dynamically replacing an objects __class__; is it safe?

[gst]

Hi,
You can subclass M2 from M1 and override only what you need. 
Regards,
-- 
https://mail.python.org/mailman/listinfo/python-list

Re: Dynamically replacing an objects __class__; is it safe?

[Steve D'Aprano]

On Wed, 15 Mar 2017 08:54 pm, marco.naw...@colosso.nl wrote:

> Dear All,
> 
> Summary of the question:
> Is it generally safe to dynamically change an objects class; if not
> under which conditions can it be considered safe.

*Generally* safe? No. You cannot expect to take an arbitrary object and
change its class:


animal = Whale()
animal.__class__ = Bird
animal.fly()


You probably can't make a whale fly just by changing the class to bird. It
will need wings, and feathers, at the very least.

For pure-Python classes, it is "safe" in the sense that the worst that will
happen is an exception, not a seg fault or core dump. But it is not safe in
the sense that the instance will work correctly: methods may fail, or do he
wrong thing.

(Actually, *silently doing the wrong thing* is usually much worse than
raising an exception or crashing. At least when the program crashes, you
know it is broken!)

For *cooperative* classes, where you design the Before and After classes to
operate correctly when you change the __class__, that's perfectly safe.


> Context:
> Given the code below, I have no direct control over Base and M1. M1
> is a instantiated by 'calling' the read-only property of Base.
> I would like to transparently switch behaviour from M1 to M2. I have
> tried several things, but finally settled on the code below. Although
> it works really well, I am not sure whether I am about to shoot myself
> in the foot. In short, what I do is derive from Base, shadow the read-only
> property 'my_prop' so it returns M2 and finally replace an objects
> __class__ attribute with my derived class.

Sounds scarily confusing, and I say that as somebody who LOVES the "swap the
__class__" trick when it is appropriate. If I wouldn't risk swapping out
the __class__ if I didn't control the classes. You have this:


o = Base()
o.my_prop
# Prints 'Initializing M1'

o = Base()
o.__class__ = ShadowBase
o.my_prop
# Prints 'Initializing M2'


But why not just do this?

o = ShadowBase()  # inherits from Base, overrides my_prop
o.my_prop
# Prints 'Initializing M2'


This sounds like an ordinary example of class inheritance to me, no need to
add complications with swapping out the __class__ for something different.


The only reason I would consider using the __class__ trick here is if you
don't control the instantiation of the object: you don't create your own
instance, you receive it pre-instantiated from elsewhere.




-- 
Steve
“Cheer up,” they said, “things could be worse.” So I cheered up, and sure
enough, things got worse.

-- 
https://mail.python.org/mailman/listinfo/python-list

Re: Dynamically replacing an objects __class__; is it safe?

[marco . nawijn]

On Wednesday, March 15, 2017 at 11:33:56 AM UTC+1, Peter Otten wrote:
> marco.naw...@colosso.nl wrote:
> 
> > Dear All,
> > 
> > Summary of the question:
> > Is it generally safe to dynamically change an objects class; if not
> > under which conditions can it be considered safe.
> > 
> > Context:
> > Given the code below, I have no direct control over Base and M1. M1
> > is a instantiated by 'calling' the read-only property of Base.
> > I would like to transparently switch behaviour from M1 to M2. I have
> > tried several things, but finally settled on the code below. Although
> > it works really well, I am not sure whether I am about to shoot myself
> > in the foot. In short, what I do is derive from Base, shadow the read-only
> > property 'my_prop' so it returns M2 and finally replace an objects
> > __class__ attribute with my derived class.
> > 
> > Any thoughts or comments?
> 
> Saying that something is "safe" is always problematic. I'd rather look for 
> ways to break the code and then decide if you can live with these potential 
> problems. I'll start with
> 
> - If Base is reimplemented in C __class__ may become read-only
> - my_prop may be accessed by Base methods and expect an M1 instance

Ok. These are valid points. The first is not a real concern to me at the
moment. The whole purpose of Base is to provide a higher level interface
to the underlying C/C++ code.

The second is more interesting. I was aware of this and checked the source
code of the library. I know an M1 instance is not directly used by Base.
In addition, I expose the exact same interface in M2. 

Thanks for the feedback. Much appreciated!

Marco
-- 
https://mail.python.org/mailman/listinfo/python-list

Re: Dynamically replacing an objects __class__; is it safe?

[Peter Otten]

marco.naw...@colosso.nl wrote:

> Dear All,
> 
> Summary of the question:
> Is it generally safe to dynamically change an objects class; if not
> under which conditions can it be considered safe.
> 
> Context:
> Given the code below, I have no direct control over Base and M1. M1
> is a instantiated by 'calling' the read-only property of Base.
> I would like to transparently switch behaviour from M1 to M2. I have
> tried several things, but finally settled on the code below. Although
> it works really well, I am not sure whether I am about to shoot myself
> in the foot. In short, what I do is derive from Base, shadow the read-only
> property 'my_prop' so it returns M2 and finally replace an objects
> __class__ attribute with my derived class.
> 
> Any thoughts or comments?

Saying that something is "safe" is always problematic. I'd rather look for 
ways to break the code and then decide if you can live with these potential 
problems. I'll start with

- If Base is reimplemented in C __class__ may become read-only
- my_prop may be accessed by Base methods and expect an M1 instance

> (code below is Python 3)
> 
> class M1(object):
> 
> def __init__(self):
> print('Initializing M1')
> 
> 
> class M2(object):
>   
>
> def __init__(self):
> print('Initializing M2')
>   
>
>   
>
> class Base(object):
>   
>
> @property
> def my_prop(self):
> return M1()
> 
> 
> class ShadowBase(Base):
> 
> @property
> def my_prop(self):
> return M2()
> 
> if __name__ == '__main__':
> 
> o = Base()
> o.my_prop
> # Prints 'Initializing M1'
> 
> o = Base()
> o.__class__ = ShadowBase
> o.my_prop
> # Prints 'Initializing M2'


-- 
https://mail.python.org/mailman/listinfo/python-list

Dynamically replacing an objects __class__; is it safe?

[marco . nawijn]

Dear All,

Summary of the question:
Is it generally safe to dynamically change an objects class; if not 
under which conditions can it be considered safe.

Context:
Given the code below, I have no direct control over Base and M1. M1
is a instantiated by 'calling' the read-only property of Base. 
I would like to transparently switch behaviour from M1 to M2. I have
tried several things, but finally settled on the code below. Although
it works really well, I am not sure whether I am about to shoot myself 
in the foot. In short, what I do is derive from Base, shadow the read-only
property 'my_prop' so it returns M2 and finally replace an objects __class__
attribute with my derived class.

Any thoughts or comments?

(code below is Python 3)

class M1(object):

def __init__(self):
print('Initializing M1')


class M2(object):

 
def __init__(self): 
 
print('Initializing M2')
 

 

 
class Base(object): 
 

 
@property   
 
def my_prop(self):  
 
return M1()


class ShadowBase(Base):

@property
def my_prop(self):
return M2()

if __name__ == '__main__':

o = Base()
o.my_prop
# Prints 'Initializing M1'

o = Base()
o.__class__ = ShadowBase
o.my_prop
# Prints 'Initializing M2'
-- 
https://mail.python.org/mailman/listinfo/python-list