summaryrefslogtreecommitdiff
path: root/class.c
diff options
context:
space:
mode:
authorJohn Hawthorn <[email protected]>2025-05-23 10:49:21 -0700
committerJohn Hawthorn <[email protected]>2025-05-29 16:02:07 -0400
commit6a62a46c3cd8a0fe3623b5363ff21ead1e755169 (patch)
treede2f13c708a66ac05db368a5a75c60b715d53450 /class.c
parent5b3f1c4c51480cbdbd1ace92b1767f14f9fff280 (diff)
Read {max_iv,variation}_count from prime classext
MAX_IV_COUNT is a hint which determines the size of variable width allocation we should use for a given class. We don't need to scope this by namespace, if we end up with larger builtin objects on some namespaces that isn't a user-visible problem, just extra memory use. Similarly variation_count is used to track if a given object has had too many branches in shapes it has used, and to use too_complex when that happens. That's also just a hint, so we can use the same value across namespaces without it being visible to users. Previously variation_count was being incremented (written to) on the RCLASS_EXT_READABLE ext, which seems incorrect if we wanted it to be different across namespaces
Notes
Notes: Merged: https://github.com/ruby/ruby/pull/13434
Diffstat (limited to 'class.c')
-rw-r--r--class.c4
1 files changed, 2 insertions, 2 deletions
diff --git a/class.c b/class.c
index 43ccb75de1..fd3276990a 100644
--- a/class.c
+++ b/class.c
@@ -338,9 +338,9 @@ rb_class_duplicate_classext(rb_classext_t *orig, VALUE klass, const rb_namespace
* * refined_class
* * as.class.allocator / as.singleton_class.attached_object
* * includer
+ * * max IV count
+ * * variation count
*/
- RCLASSEXT_MAX_IV_COUNT(ext) = RCLASSEXT_MAX_IV_COUNT(orig);
- RCLASSEXT_VARIATION_COUNT(ext) = RCLASSEXT_VARIATION_COUNT(orig);
RCLASSEXT_PERMANENT_CLASSPATH(ext) = RCLASSEXT_PERMANENT_CLASSPATH(orig);
RCLASSEXT_CLONED(ext) = RCLASSEXT_CLONED(orig);
RCLASSEXT_CLASSPATH(ext) = RCLASSEXT_CLASSPATH(orig);