Re: Printable string for 'self'

[Michael Tobis]

> What were you planning to do with this object exactly that didn't involve 
> binding it to any other names during its lifetime?

Nothing so silly as that.

The idea is not to prevent other references from binding to the object,
but to allow the object to ensure that a certain symbol always points
to it.

I no longer suspect that this is a good enough idea to redesign the
language, but it seems to come up regularly.

I will elaborate on my use case later, as I am still looking for a
satisfactory workaround. I think that someone working on a logo-like
platform wanted to be able to assign permanent names to his turtles,
which is an easier case to understand.

In both cases there is the idea to construct a platform for programmers
of intermediate skill, to expose the python environment as well as
powerful objects that they don't need to modify, but need to be able to
access by name. It would be good to be able to guarantee that the name
would be a valid reference to the object in question.

mt

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

Re: Printable string for 'self'

[Duncan Booth]

Michael Tobis wrote:

> I got in some trouble in these parts a few months back for advocating
> some sort of immutable reference, like
> 
> fred -> C("fred")
> 
> where any reassignment of the refernce during the lifetime of the
> referent would raise an exception. This seems to be seen as wrongheaded
> by greater pythonistas than myself.  I don't fully understand *why*
> this is a bad idea, but my intuitive idea that it would be very
> valuable has gone away.
> 

>>> fred = C("fred")
>>> jim = fred
RebindingError: cannot rebind  to name "jim"
>>> def showit(x):
print x


>>> showit(fred)
RebindingError: cannot rebind  to name "x"
>>> fred.amethod()
RebindingError: cannot rebind  to name "self"
>>> fred.aclassmethod()
That works!
>>> sys.getrefcount(fred)
RebindingError: cannot rebind  to name "object"
>>> del fred
>>>

What were you planning to do with this object exactly that didn't involve 
binding it to any other names during its lifetime?
-- 
http://mail.python.org/mailman/listinfo/python-list

Re: Printable string for 'self'

[Michael Tobis]

This behavior seems to be commonly wanted by people discovering Python,
and it is the rare case of something one can imagine that is really a
stretch to achieve in Python. Because more or less than one name may
refer to an object, in general an object can't know its name.

You can get part of the way there with

fred = C("fred")

and assign the string to self.name in the constructor. But you must
strictly enforce a convention in your code that the reference fred is
not reassigned to another referent for this to be of much use.

I got in some trouble in these parts a few months back for advocating
some sort of immutable reference, like

fred -> C("fred")

where any reassignment of the refernce during the lifetime of the
referent would raise an exception. This seems to be seen as wrongheaded
by greater pythonistas than myself.  I don't fully understand *why*
this is a bad idea, but my intuitive idea that it would be very
valuable has gone away.

mt

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

Re: Printable string for 'self'

[Bruno Desthuilliers]

Don Taylor a écrit :
(snip)
> My overall intent is to try to build something that can record 
> interactions against an object so that they can be replayed later for 
> testing and debugging.  I had in mind to generate the recording as a 
> sequence of Python statements.  

You may want to have a look at the command and memento patterns.
-- 
http://mail.python.org/mailman/listinfo/python-list

Re: Printable string for 'self'

[Don Taylor]

Fredrik Lundh wrote:

> Q. How can my code discover the name of an object?
> 
> A. The same way as you get the name of that cat you found on your
> porch: the cat itself cannot tell you its name, and it doesn't really
> care -- so the only way to find out what it's called is to ask all your
> neighbours if it's their cat... and don't be surprised if you'll find that
> it's known by many names, or no name at all!
> 
> from:
> 
> 
> http://www.python.org/doc/faq/programming.html#how-can-my-code-discover-the-name-of-an-object
> 
> (fwiw, the current crop of stray cats in my neighbourhood are known
> as "sune" and "the big fat red one")


I see.


"This is a project on which we have been working for the last three 
centuries since the lamasery was founded, in fact. It is somewhat alien 
to your way of thought, so I hope you will listen with an open mind 
while I explain it."

"Naturally."

"It is really quite simple. We have been compiling a list which shall 
contain all the possible names of God."
-

 From "The Nine Billion Names of God" by Arthur C. Clarke.

http://www.geocities.com/rojodos/docs/90.htm


> no, as long as you're aware that you're doing introspection, and that your
> code won't run in all Python environments.

Yes, it is introspection for a testing tool - not for production code.

Thanks,

Don.

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

Re: Printable string for 'self'

[Steve Holden]

Don Taylor wrote:
> Fredrik Lundh wrote:
> 
>>objects don't have names in Python, and the source is not part of
>>the running program.
>>
>>have you read this ?
>>
>>http://effbot.org/zone/python-objects.htm
> 
> 
> I have now.  Thank you very much.
> 
> "objects don't have names in Python": It appears from the code that 
> Michael posted that objects can discover the names that are bound to 
> themselves.  Is this true in general?  If so, then I guess it does not 
> matter which name I use as long as it is bound to the object.
> 
> "the source is not part of the running program" : Ok, but in my case I 
> would have the source that corresponds to the running program available 
> to me and the inspect module does appear to provide enough information 
> for me to find the corresponding piece of the source code.  Is there 
> something wrong with using the inspect module for this sort of work?
> 
> My overall intent is to try to build something that can record 
> interactions against an object so that they can be replayed later for 
> testing and debugging.  I had in mind to generate the recording as a 
> sequence of Python statements.  I would like to do this without 
> modifying the source of the target class.  At the moment this is just a 
> project to help me learn Python, although it would nice if it did yield 
> something useful.
> 
> Is there anything around that already does this sort of thing?
> 
Nothing springs to mind. However, given your statement of the overall 
intent, I should say that there's simply no need to record the "names" 
of the objects that are being modified.

Objects can indeed discover the names to which they are bound, albeit by 
a rather tortuous process of namespace introspection, but I don't see 
what advantages this offers in terms of how a given program modifies its 
environment (or namespace). Surely the important thing is that objects 
are created, modified in various ways, and finally destroyed?

If you are hoping to record "interactions against an object" without 
modifying the object's implementation I suspect your efforts may be 
doomed to failure. This opinion is offered mostly to motivate the many 
wrong-headed and knowledgeable individuals who would gain pleasure from 
proving my assertion incorrect and thereby helping you to solve your 
problem. Good luck.

regards
  Steve
-- 
Steve Holden   +44 150 684 7255  +1 800 494 3119
Holden Web LLC/Ltd www.holdenweb.com
Love me, love my blog holdenweb.blogspot.com

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

Re: Printable string for 'self'

[Fredrik Lundh]

Don Taylor wrote:

> Fredrik Lundh wrote:
> > objects don't have names in Python, and the source is not part of
> > the running program.
> >
> > have you read this ?
> >
> > http://effbot.org/zone/python-objects.htm
>
> I have now.  Thank you very much.
>
> "objects don't have names in Python": It appears from the code that
> Michael posted that objects can discover the names that are bound to
> themselves.  Is this true in general?

to the extent they have names, and the names exist in a namespace that
you can reach.  it might be time to quote myself here:

Q. How can my code discover the name of an object?

A. The same way as you get the name of that cat you found on your
porch: the cat itself cannot tell you its name, and it doesn't really
care -- so the only way to find out what it's called is to ask all your
neighbours if it's their cat... and don't be surprised if you'll find that
it's known by many names, or no name at all!

from:


http://www.python.org/doc/faq/programming.html#how-can-my-code-discover-the-name-of-an-object

(fwiw, the current crop of stray cats in my neighbourhood are known
as "sune" and "the big fat red one")

> "the source is not part of the running program" : Ok, but in my case I
> would have the source that corresponds to the running program available
> to me and the inspect module does appear to provide enough information
> for me to find the corresponding piece of the source code.  Is there
> something wrong with using the inspect module for this sort of work?

no, as long as you're aware that you're doing introspection, and that your
code won't run in all Python environments.





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

Re: Printable string for 'self'

[Don Taylor]

Fredrik Lundh wrote:
> objects don't have names in Python, and the source is not part of
> the running program.
> 
> have you read this ?
> 
> http://effbot.org/zone/python-objects.htm

I have now.  Thank you very much.

"objects don't have names in Python": It appears from the code that 
Michael posted that objects can discover the names that are bound to 
themselves.  Is this true in general?  If so, then I guess it does not 
matter which name I use as long as it is bound to the object.

"the source is not part of the running program" : Ok, but in my case I 
would have the source that corresponds to the running program available 
to me and the inspect module does appear to provide enough information 
for me to find the corresponding piece of the source code.  Is there 
something wrong with using the inspect module for this sort of work?

My overall intent is to try to build something that can record 
interactions against an object so that they can be replayed later for 
testing and debugging.  I had in mind to generate the recording as a 
sequence of Python statements.  I would like to do this without 
modifying the source of the target class.  At the moment this is just a 
project to help me learn Python, although it would nice if it did yield 
something useful.

Is there anything around that already does this sort of thing?

Cheers,

Don.

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

Re: Printable string for 'self'

[Fredrik Lundh]

Don Taylor wrote:

> And if I call my method: A().qualify global_names_bound_to_me() then I
> get an empty list back.
>
> Is what I want to do not possible without referring back to the source?

objects don't have names in Python, and the source is not part of
the running program.

have you read this ?

http://effbot.org/zone/python-objects.htm





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

Re: Printable string for 'self'

[Ben Cartwright]

Don Taylor wrote:
> Is there a way to discover the original string form of the instance that
> is represented by self in a method?
>
> For example, if I have:
>
>   fred = C()
>   fred.meth(27)
>
> then I would like meth to be able to print something like:
>
>   about to call meth(fred, 27) or
>   about to call fred.meth(27)
>
> instead of:
>
>   about to call meth(<__main__.C instance at 0x00A9D238>, 27)


Not a direct answer to your question, but this may be what you want:
If you give class C a __repr__ method you can avoid the default  string.

  >>> class C(object):
  def __init__(self, name):
  self.name = name
  def __repr__(self):
  return '%s(%r)' % (self.__class__.__name__, self.name)
  def meth(self, y):
  print 'about to call %r.meth(%r)' % (self, y)

  >>> fred = C('Fred')
  >>> fred.meth(27)
  about to call C('Fred').meth(27)
  >>> def meth2(x, y):
  print 'about to call meth2(%r, %r)' % (x, y)

  >>> meth2(fred, 42)
  about to call meth2(C('Fred'), 42)


Of course, this doesn't tell you the name of the variable that points
to the instance.  For that (here's your direct answer), you will have
to go to the source:

  >>> import inspect
  >>> import linecache
  >>> def f(a, b):
  print 'function call from parent scope:'
  caller = inspect.currentframe().f_back
  filename = caller.f_code.co_filename
  linecache.checkcache(filename)
  line = linecache.getline(filename, caller.f_lineno)
  print '' + line.strip()
  return a*b

  >>> fred = 4
  >>> x = f(3, fred)
  function call from parent scope:
  x = f(3, fred)


But defining __repr__ is far easier and more common.

--Ben

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

Re: Printable string for 'self'

[Terry Reedy]

"Don Taylor" <[EMAIL PROTECTED]> wrote in message 
news:[EMAIL PROTECTED]
> Is there a way to discover the original string form of the instance that
> is represented by self in a method?
> For example, if I have:
> fred = C()
> fred.meth(27)
> then I would like meth to be able to print something like:
> about to call meth(fred, 27) or
> about to call fred.meth(27)
> instead of:
> about to call meth(<__main__.C instance at 0x00A9D238>, 27)

That last *is* the default 'original string form of the instance'. 
Consider
>>> [C() for i in [1,2,3]]
[<__main__.C object at 0x00B536F0>, <__main__.C object at 0x00B53D90>, 
<__main__.C object at 0x00B53830>]

If you want your instances to have a name attribute, as with functions, 
classes, and modules, then give them one and supply it to the init method. 
Note that it does not have to be a bindable name, just a string.  IE, 
'a[1]' could be the string identifier.

Terry Jan Reedy



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

Re: Printable string for 'self'

[Don Taylor]

Michael Spencer wrote:
>
> 
> In general, this requires exhaustive search of name bindings e.g.,:
> 
>  >>> def get_names_of(obj, ns):
> ... return [name for name, value in ns.iteritems() if value is obj]
> ...
>  >>> class A(object):
> ... def global_names_bound_to_me(self):
> ... return get_names_of(self, globals())
> ...
>  >>> a = A()
>  >>> a.global_names_bound_to_me()
> ['a']
>  >>> b = a
>  >>> a.global_names_bound_to_me()
> ['a', 'b']
>  >>>
> 
Ah, ok.

But, as you show in the example, this technique does not let me figure 
out which of 'a' or 'b' was used to qualify global_names_bound_to_me().

And if I call my method: A().qualify global_names_bound_to_me() then I 
get an empty list back.

Is what I want to do not possible without referring back to the source?

Don.


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

Re: Printable string for 'self'

[Michael Spencer]

Don Taylor wrote:
> Is there a way to discover the original string form of the instance that 
> is represented by self in a method?
> 
> For example, if I have:
> 
>   fred = C()
>   fred.meth(27)
> 
> then I would like meth to be able to print something like:
> 
>   about to call meth(fred, 27) or
>   about to call fred.meth(27)
> 
> instead of:
> 
>   about to call meth(<__main__.C instance at 0x00A9D238>, 27)
> 
> Thanks in advance,
> 
> Don.
> 
In general, this requires exhaustive search of name bindings e.g.,:

 >>> def get_names_of(obj, ns):
... return [name for name, value in ns.iteritems() if value is obj]
...
 >>> class A(object):
... def global_names_bound_to_me(self):
... return get_names_of(self, globals())
...
 >>> a = A()
 >>> a.global_names_bound_to_me()
['a']
 >>> b = a
 >>> a.global_names_bound_to_me()
['a', 'b']
 >>>

HTH
Michael

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

Printable string for 'self'

[Don Taylor]

Is there a way to discover the original string form of the instance that 
is represented by self in a method?

For example, if I have:

fred = C()
fred.meth(27)

then I would like meth to be able to print something like:

about to call meth(fred, 27) or
about to call fred.meth(27)

instead of:

about to call meth(<__main__.C instance at 0x00A9D238>, 27)

Thanks in advance,

Don.

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