[pypy-commit] pypy default: issue #1522

[cfbolz]

Author: Carl Friedrich Bolz <cfbolz at gmx.de>
Branch: 
Changeset: r80492:fbabd7db9dad
Date: 2015-10-31 13:27 +0100
http://bitbucket.org/pypy/pypy/changeset/fbabd7db9dad/

Log:	issue #1522

	this fixes the problem of making a tiny set out of a really huge
	iterable without generating a list out of the iterable first.

diff --git a/pypy/objspace/std/setobject.py b/pypy/objspace/std/setobject.py
--- a/pypy/objspace/std/setobject.py
+++ b/pypy/objspace/std/setobject.py
@@ -1589,18 +1589,18 @@
         w_set.sstorage = strategy.get_storage_from_unwrapped_list(intlist)
         return
 
+    length_hint = space.length_hint(w_iterable, 0)
+
+    if jit.isconstant(length_hint):
+        return _pick_correct_strategy_unroll(space, w_set, w_iterable)
+
+    _create_from_iterable(space, w_set, w_iterable)
+
+
+ at jit.unroll_safe
+def _pick_correct_strategy_unroll(space, w_set, w_iterable):
+
     iterable_w = space.listview(w_iterable)
-
-    if len(iterable_w) == 0:
-        w_set.strategy = strategy = space.fromcache(EmptySetStrategy)
-        w_set.sstorage = strategy.get_empty_storage()
-        return
-
-    _pick_correct_strategy(space, w_set, iterable_w)
-
- at jit.look_inside_iff(lambda space, w_set, iterable_w:
-        jit.loop_unrolling_heuristic(iterable_w, len(iterable_w), UNROLL_CUTOFF))
-def _pick_correct_strategy(space, w_set, iterable_w):
     # check for integers
     for w_item in iterable_w:
         if type(w_item) is not W_IntObject:
@@ -1640,6 +1640,23 @@
     w_set.strategy = space.fromcache(ObjectSetStrategy)
     w_set.sstorage = w_set.strategy.get_storage_from_list(iterable_w)
 
+
+create_set_driver = jit.JitDriver(name='create_set',
+                                  greens=['tp', 'strategy'],
+                                  reds='auto')
+
+def _create_from_iterable(space, w_set, w_iterable):
+    iterable = space.iteriterable(w_iterable)
+    w_set.strategy = strategy = space.fromcache(EmptySetStrategy)
+    w_set.sstorage = strategy.get_empty_storage()
+
+    tp = space.type(w_iterable)
+    for w_item in space.iteriterable(w_iterable):
+        create_set_driver.jit_merge_point(tp=tp, strategy=w_set.strategy)
+        w_set.add(w_item)
+
+
+
 init_signature = Signature(['some_iterable'], None, None)
 init_defaults = [None]
 def _initialize_set(space, w_obj, w_iterable=None):