Acctually, the comment in gencgc.c, which mentions Bruno Haible's write-up, seems to be too optimistic. Currently cost of weak hash table GCing is O(W^2*N), not O(W^2+N).
Acctually, the comment in gencgc.c, which mentions Bruno Haible's write-up, seems to be too optimistic.
Currently cost of weak hash table GCing is O(W^2*N), not O(W^2+N).