SF patch 576101, by Oren Tirosh: alternative implementation of

interning.  I modified Oren's patch significantly, but the basic idea
and most of the implementation is unchanged.  Interned strings created
with PyString_InternInPlace() are now mortal, and you must keep a
reference to the resulting string around; use the new function
PyString_InternImmortal() to create immortal interned strings.

2 files changed, 15 insertions, 4 deletions

diff --git a/Include/modsupport.h b/Include/modsupport.h
index 2e577ab..7e69921 100644
--- a/Include/modsupport.h
+++ b/Include/modsupport.h
@@ -23,8 +23,8 @@ PyAPI_FUNC(int) PyModule_AddObject(PyObject *, char *, PyObject *);
 PyAPI_FUNC(int) PyModule_AddIntConstant(PyObject *, char *, long);
 PyAPI_FUNC(int) PyModule_AddStringConstant(PyObject *, char *, char *);
 
-#define PYTHON_API_VERSION 1011
-#define PYTHON_API_STRING "1011"
+#define PYTHON_API_VERSION 1012
+#define PYTHON_API_STRING "1012"
 /* The API version is maintained (independently from the Python version)
    so we can detect mismatches between the interpreter and dynamically
    loaded modules.  These are diagnosed by an error message but
@@ -38,6 +38,9 @@ PyAPI_FUNC(int) PyModule_AddStringConstant(PyObject *, char *, char *);
    Please add a line or two to the top of this log for each API
    version change:
 
+   19-Aug-2002  GvR	1012	Changes to string object struct for
+   				interning changes, saving 3 bytes.
+
    17-Jul-2001	GvR	1011	Descr-branch, just to be on the safe side
 
    25-Jan-2001  FLD     1010    Parameters added to PyCode_New() and
diff --git a/Include/stringobject.h b/Include/stringobject.h
index fd0f49a..ff30704 100644
--- a/Include/stringobject.h
+++ b/Include/stringobject.h
@@ -25,7 +25,7 @@ functions should be applied to nil objects.
 */
 
 /* Caching the hash (ob_shash) saves recalculation of a string's hash value.
-   Interning strings (ob_sinterned) tries to ensure that only one string
+   Interning strings (ob_sstate) tries to ensure that only one string
    object with a given value exists, so equality tests can be one pointer
    comparison.  This is generally restricted to strings that "look like"
    Python identifiers, although the intern() builtin can be used to force
@@ -35,10 +35,14 @@ functions should be applied to nil objects.
 typedef struct {
     PyObject_VAR_HEAD
     long ob_shash;
-    PyObject *ob_sinterned;
+    int ob_sstate;
     char ob_sval[1];
 } PyStringObject;
 
+#define SSTATE_NOT_INTERNED 0
+#define SSTATE_INTERNED_MORTAL 1
+#define SSTATE_INTERNED_IMMORTAL 2
+
 PyAPI_DATA(PyTypeObject) PyBaseString_Type;
 PyAPI_DATA(PyTypeObject) PyString_Type;
 
@@ -66,9 +70,13 @@ extern DL_IMPORT(PyObject *) PyString_DecodeEscape(const char *, int,
 						   const char *);
 
 PyAPI_FUNC(void) PyString_InternInPlace(PyObject **);
+PyAPI_FUNC(void) PyString_InternImmortal(PyObject **);
 PyAPI_FUNC(PyObject *) PyString_InternFromString(const char *);
 PyAPI_FUNC(void) _Py_ReleaseInternedStrings(void);
 
+/* Use only if you know it's a string */
+#define PyString_CHECK_INTERNED(op) (((PyStringObject *)(op))->ob_sstate)
+
 /* Macro, trading safety for speed */
 #define PyString_AS_STRING(op) (((PyStringObject *)(op))->ob_sval)
 #define PyString_GET_SIZE(op)  (((PyStringObject *)(op))->ob_size)