You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We are thinking about getting rid of OMR_GC_LEAF_BITS as an optional feature, and making it permanent. This would remove some duplicated code in our object scanners which is never exercised.
To support people who don't use leaf maps, we could add default implementation of ObjectScanner::getNextSlotMap(uintptr_t *slotMap, uintptr_t *leafMap, bool *hasNextSlotMap) that defaults to setting the leaf map to 0. This would mean that the API stays the same. It might have some effect on performance for those who don't use leaf maps. I'm thinking something like:
We are thinking about getting rid of
OMR_GC_LEAF_BITS
as an optional feature, and making it permanent. This would remove some duplicated code in our object scanners which is never exercised.To support people who don't use leaf maps, we could add default implementation of
ObjectScanner::getNextSlotMap(uintptr_t *slotMap, uintptr_t *leafMap, bool *hasNextSlotMap)
that defaults to setting the leaf map to 0. This would mean that the API stays the same. It might have some effect on performance for those who don't use leaf maps. I'm thinking something like:The alternatives are we leave it as an optional feature, or we force users to always specify *leafMap = 0 when they don't support leaf maps (edited)
The text was updated successfully, but these errors were encountered: