When C code needs to compare Python object with int 0 or add int 1 it either use local reference to PyLong_FromLong(0) and PyLong_FromLong(1) which should be decrefed just after use or module level global variable initialized and cleared during initializing and finalizing the module.

Proposed patch adds global variables _PyLong_Zero and _PyLong_One for references to integer objects 0 and 1. This simplifies the code since no need to initialize local variables, check for error the result of PyLong_FromLong() and decref it after use. The patch decreases the total code size by 244 lines.

That variables are only for internal use. User code should use PyLong_FromLong(0) and PyLong_FromLong(1).

I like it.

+1. Please create a PR for it.

+1 for this idea.

Also consider adding new function PyLong_Increment. This basic operation is small pain using the current API. It may also give a small speed benefit.

Other common values used in C functions: empty byte string, empty Unicode string, empty tuple. The problem is to make sure that singletons are created in the right order :-/

This issue reminded me an old idea of writing a generalization of the _Py_IDENTIFIER() API. I created an issue bpo-29881: Add a new private API for "static C variables" (_PyStaticVar) to clear them at exit. To be clear: it's related but different to this issue, the two issues are exclusive.

Yet one idea that can make the code simpler is make PyLong_FromLong(0) and PyLong_FromLong(1) never failing. I.e. require NSMALLPOSINTS not less than 2.

Also consider adding new function PyLong_Increment. This basic operation is small pain using the current API. It may also give a small speed benefit.

Smaller pain with using _PyLong_One and Py_SETREF().

    Py_SETREF(long_obj, PyNumber_Add(long_obj, _PyLong_One));

Agree that with _PyLong_Increment() it can look better and be faster. But I don't know whether incrementing by 1 is enough popular operation. I have counted 5 cases in the stdlib (not counting tests): for enumerate, range and Counter.

The problem is to make sure that singletons are created in the right order :-/

Yes, I spent much time for making empty Unicode string singleton always be initialized. It can be accessed at very early stage.

New changeset ba85d69 by Serhiy Storchaka in branch 'master':
bpo-29878: Add global instances of int for 0 and 1. (#852)
ba85d69

New changeset bae6881 by Serhiy Storchaka in branch 'master':
Update Argument Clinic generated code for bpo-29878. (bpo-1001)
bae6881