Marginal performance improvement in LockBuffer --- calculate address
of BufferLocks[] entry just once. Seems to save 10% or so of the routine's runtime, which'd not be worth worrying about if it weren't such a hotspot.
Showing
Please register or sign in to comment