What determines which strings are interned and when? [duplicate]

Question

>>> s1 = "spam"
>>> s2 = "spam"
>>> s1 is s2
True
>>> q = 'asdalksdjfla;ksdjf;laksdjfals;kdfjasl;fjasdf'
>>> r = 'asdalksdjfla;ksdjf;laksdjfals;kdfjasl;fjasdf'
>>> q is r
False

How many characters should have to s1 is s2 give False? Where is limit? i.e. I am asking how long a string has to be before python starts making separate copies of it.

Spurn

Are you asking how long a string has to be before python starts making separate copies of it? – Buller

It's completely implementation dependent and could change in the next version of Python without warning. Why do you want to know? – Cracker

Yes, I am asking how long a string has to be before python starts making separate copies of it. – Spurn

@Cris: What are you trying to accomplish? The behavior is implementation-defined and cannot be relied on. Why not just check for equality? – Wigwag

I just want know for knowing. Don't know that this is no important and very flexible. – Spurn

if you just want to know, why don't you test it on the implementation you're using? – Benedick

if i will only testing i will never get knowledge that this is implementation dependent – Spurn

Note that you may get a different result if you put your example statements in a script, instead of trying them interactively. – Crucial

Answer 1

String interning is implementation specific and shouldn't be relied upon, use equality testing if you want to check two strings are identical.

Georgeta

...implementation specific. This is also answer. Thank You. – Spurn

@Cris As a concrete example, ideone's python version seems to intern your "long" strings as well as 1000-character-eval-constructed strings. – Shunt

It gives me little more knowlednge and many more questions, byt it is ok. – Spurn

So is there never a legit use case for users to test s1 is s2 ? – Cassaundracassava

Answer 2

If you want, for some bizarre reason, to force the comparison to be true then use the intern function:

>>> a = intern('12345678012345678901234567890qazwsxedcrfvtgbyhnujmikolp')
>>> b = intern('12345678012345678901234567890qazwsxedcrfvtgbyhnujmikolp')
>>> a is b
True

Invalidate

For some bizarre reason, good to know. – Spurn

Additionally, intern is moved into module sys in python 3000 along with id. Also good to know. – Invalidate

For some reasons, the most. Could You tell is intern() only python 3000 toy? – Spurn

@Cris intern() has been in Python at least since version 1.5 – Shunt

intern is in the root namespace. – Invalidate

It is built-in, you don't need to import anything. – Mussel

Answer 3

Here is a piece of comment about interned string from CPython 2.5.0 source file (stringobject.h)

/* ... ... This is generally restricted to strings that **"look like" Python identifiers**, although the intern() builtin can be used to force interning of any string ... ... */

Accordingly, strings contain only underscores, digits or alphabets will be interned. In your example, q and ``r contain ;, so they will not be interned.

Streptothricin