obsolete: use C code for headrevs calculation
authorDurham Goode <durham@fb.com>
Tue, 16 Sep 2014 16:03:21 -0700
changeset 22484 2b5940f64750
parent 22483 a21d04848359
child 22485 efedda4aed49
obsolete: use C code for headrevs calculation Previously, if there were filtered revs the repository could not use the C fast path for computing the head revs in the changelog. This slowed down many operations in large repositories. This adds the ability to filter revs to the C fast path. This speeds up histedit on repositories with filtered revs by 30% (13s to 9s). This could be improved further by sorting the filtered revs and walking the sorted list while we walk the changelog, but even this initial version that just calls __contains__ is still massively faster. The new C api is compatible for both new and old python clients, and the new python client can call both new and old C apis.
mercurial/changelog.py
mercurial/parsers.c
--- a/mercurial/changelog.py	Tue Sep 16 23:47:34 2014 -0700
+++ b/mercurial/changelog.py	Tue Sep 16 16:03:21 2014 -0700
@@ -171,8 +171,13 @@
 
     def headrevs(self):
         if self.filteredrevs:
-            # XXX we should fix and use the C version
-            return self._headrevs()
+            try:
+                return self.index.headrevs(self.filteredrevs)
+            # AttributeError covers non-c-extension environments.
+            # TypeError allows us work with old c extensions.
+            except (AttributeError, TypeError):
+                return self._headrevs()
+
         return super(changelog, self).headrevs()
 
     def strip(self, *args, **kwargs):
--- a/mercurial/parsers.c	Tue Sep 16 23:47:34 2014 -0700
+++ b/mercurial/parsers.c	Tue Sep 16 16:03:21 2014 -0700
@@ -508,6 +508,7 @@
 	Py_ssize_t length;     /* current number of elements */
 	PyObject *added;       /* populated on demand */
 	PyObject *headrevs;    /* cache, invalidated on changes */
+	PyObject *filteredrevs;/* filtered revs set */
 	nodetree *nt;          /* base-16 trie */
 	int ntlength;          /* # nodes in use */
 	int ntcapacity;        /* # nodes allocated */
@@ -823,15 +824,60 @@
 	return newlist;
 }
 
-static PyObject *index_headrevs(indexObject *self)
+static int check_filter(PyObject *filter, Py_ssize_t arg) {
+	if (filter) {
+		PyObject *arglist, *result;
+		int isfiltered;
+
+		arglist = Py_BuildValue("(n)", arg);
+		if (!arglist) {
+			return -1;
+		}
+
+		result = PyEval_CallObject(filter, arglist);
+		Py_DECREF(arglist);
+		if (!result) {
+			return -1;
+		}
+
+		/* PyObject_IsTrue returns 1 if true, 0 if false, -1 if error,
+		 * same as this function, so we can just return it directly.*/
+		isfiltered = PyObject_IsTrue(result);
+		Py_DECREF(result);
+		return isfiltered;
+	} else {
+		return 0;
+	}
+}
+
+static PyObject *index_headrevs(indexObject *self, PyObject *args)
 {
 	Py_ssize_t i, len, addlen;
 	char *nothead = NULL;
 	PyObject *heads;
+	PyObject *filter = NULL;
+	PyObject *filteredrevs = Py_None;
 
-	if (self->headrevs)
+	if (!PyArg_ParseTuple(args, "|O", &filteredrevs)) {
+		return NULL;
+	}
+
+	if (self->headrevs && filteredrevs == self->filteredrevs)
 		return list_copy(self->headrevs);
 
+	Py_DECREF(self->filteredrevs);
+	self->filteredrevs = filteredrevs;
+	Py_INCREF(filteredrevs);
+
+	if (filteredrevs != Py_None) {
+		filter = PyObject_GetAttrString(filteredrevs, "__contains__");
+		if (!filter) {
+			PyErr_SetString(PyExc_TypeError,
+				"filteredrevs has no attribute __contains__");
+			goto bail;
+		}
+	}
+
 	len = index_length(self) - 1;
 	heads = PyList_New(0);
 	if (heads == NULL)
@@ -850,9 +896,25 @@
 		goto bail;
 
 	for (i = 0; i < self->raw_length; i++) {
-		const char *data = index_deref(self, i);
-		int parent_1 = getbe32(data + 24);
-		int parent_2 = getbe32(data + 28);
+		const char *data;
+		int parent_1, parent_2, isfiltered;
+
+		isfiltered = check_filter(filter, i);
+		if (isfiltered == -1) {
+			PyErr_SetString(PyExc_TypeError,
+				"unable to check filter");
+			goto bail;
+		}
+
+		if (isfiltered) {
+			nothead[i] = 1;
+			continue;
+		}
+
+		data = index_deref(self, i);
+		parent_1 = getbe32(data + 24);
+		parent_2 = getbe32(data + 28);
+
 		if (parent_1 >= 0)
 			nothead[parent_1] = 1;
 		if (parent_2 >= 0)
@@ -866,12 +928,26 @@
 		PyObject *p1 = PyTuple_GET_ITEM(rev, 5);
 		PyObject *p2 = PyTuple_GET_ITEM(rev, 6);
 		long parent_1, parent_2;
+		int isfiltered;
 
 		if (!PyInt_Check(p1) || !PyInt_Check(p2)) {
 			PyErr_SetString(PyExc_TypeError,
 					"revlog parents are invalid");
 			goto bail;
 		}
+
+		isfiltered = check_filter(filter, i);
+		if (isfiltered == -1) {
+			PyErr_SetString(PyExc_TypeError,
+				"unable to check filter");
+			goto bail;
+		}
+
+		if (isfiltered) {
+			nothead[i] = 1;
+			continue;
+		}
+
 		parent_1 = PyInt_AS_LONG(p1);
 		parent_2 = PyInt_AS_LONG(p2);
 		if (parent_1 >= 0)
@@ -894,9 +970,11 @@
 
 done:
 	self->headrevs = heads;
+	Py_XDECREF(filter);
 	free(nothead);
 	return list_copy(self->headrevs);
 bail:
+	Py_XDECREF(filter);
 	Py_XDECREF(heads);
 	free(nothead);
 	return NULL;
@@ -1896,6 +1974,8 @@
 	self->cache = NULL;
 	self->data = NULL;
 	self->headrevs = NULL;
+	self->filteredrevs = Py_None;
+	Py_INCREF(Py_None);
 	self->nt = NULL;
 	self->offsets = NULL;
 
@@ -1945,6 +2025,7 @@
 static void index_dealloc(indexObject *self)
 {
 	_index_clearcaches(self);
+	Py_XDECREF(self->filteredrevs);
 	Py_XDECREF(self->data);
 	Py_XDECREF(self->added);
 	PyObject_Del(self);
@@ -1977,7 +2058,7 @@
 	 "clear the index caches"},
 	{"get", (PyCFunction)index_m_get, METH_VARARGS,
 	 "get an index entry"},
-	{"headrevs", (PyCFunction)index_headrevs, METH_NOARGS,
+	{"headrevs", (PyCFunction)index_headrevs, METH_VARARGS,
 	 "get head revisions"},
 	{"insert", (PyCFunction)index_insert, METH_VARARGS,
 	 "insert an index entry"},