lru_cache: Add more comments. Fix comment typos. Clarify a comment. (GH-11795)

This commit is contained in:
Raymond Hettinger 2019-02-08 18:55:02 -08:00 committed by Miss Islington (bot)
parent 7ab3d1573c
commit 2dda72a2e8
1 changed files with 25 additions and 4 deletions

View File

@ -661,6 +661,26 @@ sequence is empty.");
/* lru_cache object **********************************************************/ /* lru_cache object **********************************************************/
/* There are four principal algorithmic differences from the pure python version:
1). The C version relies on the GIL instead of having its own reentrant lock.
2). The prev/next link fields use borrowed references.
3). For a full cache, the pure python version rotates the location of the
root entry so that it never has to move individual links and it can
limit updates to just the key and result fields. However, in the C
version, links are temporarily removed while the cache dict updates are
occurring. Afterwards, they are appended or prepended back into the
doubly-linked lists.
4) In the Python version, the _HashSeq class is used to prevent __hash__
from being called more than once. In the C version, the "known hash"
variants of dictionary calls as used to the same effect.
*/
/* this object is used delimit args and keywords in the cache keys */ /* this object is used delimit args and keywords in the cache keys */
static PyObject *kwd_mark = NULL; static PyObject *kwd_mark = NULL;
@ -1009,14 +1029,15 @@ bounded_lru_cache_wrapper(lru_cache_object *self, PyObject *args, PyObject *kwds
link = self->root.next; link = self->root.next;
lru_cache_extract_link(link); lru_cache_extract_link(link);
/* Remove it from the cache. /* Remove it from the cache.
The cache dict holds one reference to the link, The cache dict holds one reference to the link.
and the linked list holds yet one reference to it. */ We created one other reference when the link was created.
The linked list only has borrowed references. */
popresult = _PyDict_Pop_KnownHash(self->cache, link->key, popresult = _PyDict_Pop_KnownHash(self->cache, link->key,
link->hash, Py_None); link->hash, Py_None);
if (popresult == Py_None) { if (popresult == Py_None) {
/* Getting here means that the user function call or another /* Getting here means that the user function call or another
thread has already removed the old key from the dictionary. thread has already removed the old key from the dictionary.
This link is now an orpan. Since we don't want to leave the This link is now an orphan. Since we don't want to leave the
cache in an inconsistent state, we don't restore the link. */ cache in an inconsistent state, we don't restore the link. */
Py_DECREF(popresult); Py_DECREF(popresult);
Py_DECREF(link); Py_DECREF(link);
@ -1048,7 +1069,7 @@ bounded_lru_cache_wrapper(lru_cache_object *self, PyObject *args, PyObject *kwds
prev and next fields set to valid values. We have to wait prev and next fields set to valid values. We have to wait
for successful insertion in the cache dict before adding the for successful insertion in the cache dict before adding the
link to the linked list. Otherwise, the potentially reentrant link to the linked list. Otherwise, the potentially reentrant
__eq__ call could cause the then ophan link to be visited. */ __eq__ call could cause the then orphan link to be visited. */
if (_PyDict_SetItem_KnownHash(self->cache, key, (PyObject *)link, if (_PyDict_SetItem_KnownHash(self->cache, key, (PyObject *)link,
hash) < 0) { hash) < 0) {
/* Somehow the cache dict update failed. We no longer can /* Somehow the cache dict update failed. We no longer can