LuaSnap is now working based on eris (at least superficially).

luprex/core/cpp/luasnap.hpp

@@ -6,17 +6,8 @@
 // possible to roll the entire interpreter back to a previously-snapshotted
 // state.
 //
-// To accomplish this, we take advantage of the 'allocf' parameter to
-// lua_newstate.  This lets us hook the lua allocator, which in turn lets us
-// find every block of memory currently in use by lua.  To snapshot, we 'memcpy'
-// every block of memory that lua is using into a buffer.  To rollback, we just
-// 'memcpy' the data back.
-//
-// The current implementation is a proof-of-concept.  It's quite wasteful of
-// memory, roughly doubling the amount of RAM that LUA uses.  It's also not
-// super-fast, since it allocates tons of tiny blocks.  But it totally
-// works, so it demonstrate that this method of snapshot and rollback is
-// feasible.
+// To accomplish this, we use eris serialization.  This is messy and not
+// very modular, but it does work.
 //
 /////////////////////////////////////////////////////////////////////////////