Re: negative numbers are not equal...

[ariel ledesma]

Christian Heimes wrote:

You are getting the result because Python optimized small integers.
See http://svn.python.org/projects/pytho...ts/intobject.c
Integers between -5 and +256 are singletons as are some other objects
like strings with one element or empty tuples. You must not rely on
the optimization.

i see now, so i guess that's also why id() returns the same address for
them as well...
i'll dive into the implementation file, it may be a bit out of my league
but i'll see what i can gather, and hey, that's how it works, right? :-)

thanks

ariel

[Wojtek Walczak]

On Thu, 14 Aug 2008 18:23:21 -0300, ariel ledesma wrote:

i see now, so i guess that's also why id() returns the same address for
them as well...

It just have to work like this.

a is b

is actually equal to:

id(a) == id(b)

so there is no other way for id() in such case.

Hope this helps.

--
Regards,
Wojtek Walczak,
http://www.stud.umk.pl/~wojtekwa/

[castironpi]

On Aug 14, 4:31*pm, Wojtek Walczak <[e-mail]:

On Thu, 14 Aug 2008 18:23:21 -0300, ariel ledesma wrote:

i see now, so i guess that's also why id() returns the same address for
them as well...

It just have to work like this.

a is b

is actually equal to:

id(a) == id(b)

so there is no other way for id() in such case.

Hope this helps.

--
Regards,
Wojtek Walczak,http://www.stud.umk.pl/~wojtekwa/

For

a= 6
b= a

the test

a is b

should clearly return true. Python distinguishes what mathematics
does not, between identity and equality. Clearly 5+4 and 6+3 -
evaluate- to the same, but math doesn't define whether they are the
same, and in some sense the question isn't asked ordinarily, or isn't
debated. I want to infer that math doesn't define the 'is' relation
as Python knows it.

I feel the documentation should state, 'the interpreter is free to
return a -new- equivalent non-identical object in the case of
immutables.'

My tests:

>>a= -6
a is -6

False

>>-6 is -6

True

I don't know a convincing argument for the truth of Is( -6, -6 ).
Perhaps you could make one, or one for the permissibility of Is( a,
b ) & ~Equal( a, b )... identical non-equivalent.

[Steven D'Aprano]

On Thu, 14 Aug 2008 20:44:32 -0700, castironpi wrote:

For

a= 6
b= a

the test

a is b

should clearly return true.

Since Python promises not to make a copy of a when you execute "b = a",
then I think that such behaviour is guaranteed by the language.

Python distinguishes what mathematics does
not, between identity and equality. Clearly 5+4 and 6+3 - evaluate- to
the same, but math doesn't define whether they are the same, and in some
sense the question isn't asked ordinarily, or isn't debated. I want to
infer that math doesn't define the 'is' relation as Python knows it.

Mathematicians often *define* equality as identity. That certainly makes
sense when dealing with numbers -- what would it mean to say that there
are (say) three different instances of the abstract integer 42, all equal
yet not identical? I suggest that this simply doesn't make sense -- it is
"not even wrong".

Equality-as-identity may not hold in all areas of mathematics, but I
think it is safe to say it holds for ideal (abstract) numbers, as opposed
to implementations of numbers as bit patterns or objects in memory.

http://en.wikipedia.org/wiki/Equality_(mathematics)

--
Steven

[ariel ledesma]

well, i'm glad i stumbled upon this detail early on (i only had to fix
about one page of code)... i'll just stick to 'is' when it concerns
checking if it is the *same* object (memorywise) instead of an
*equivalent* one...
just before wrapping up, the special methods __eq__ and __ne__ are
called upon == and !=, right? not for 'is', 'is not'...

Steven D'Aprano wrote:

>
Mathematicians often *define* equality as identity. That certainly makes
sense when dealing with numbers -- what would it mean to say that there
are (say) three different instances of the abstract integer 42, all equal
yet not identical? I suggest that this simply doesn't make sense -- it is
"not even wrong".

Equality-as-identity may not hold in all areas of mathematics, but I
think it is safe to say it holds for ideal (abstract) numbers, as opposed
to implementations of numbers as bit patterns or objects in memory.

http://en.wikipedia.org/wiki/Equality_(mathematics)

but who knows? maybe abstract numbers -5 to 256 are optimized as well! :-)

[Christian Heimes]

ariel ledesma wrote:

well, i'm glad i stumbled upon this detail early on (i only had to fix
about one page of code)... i'll just stick to 'is' when it concerns
checking if it is the *same* object (memorywise) instead of an
*equivalent* one...
just before wrapping up, the special methods __eq__ and __ne__ are
called upon == and !=, right? not for 'is', 'is not'...

Correct! You can't change the identity operator 'is' with a magic method.

Christian