Reply to thread

Message: <blockquote data-quote="deaf_judger" data-source="post: 68202" data-attributes="member: 390">Not in the same execution pass.Yes. During evaluation I store: AST, Values, Lists of `Effect`, Hashes, and our dependency graph, and during execution we have our ledger entries in sqlite and  our dependency resolution state. This is heavy, but not absurdly heavy because 1000 identical `(print "hello")` will all hash to the same ID. The memory growth depends on structural uniqueness.During evaluation: The program produces values. Some of those values are Effect descriptions. At the end, the runtime walks the result graph and extracts them. Only the final value graph, the set of reachable effects, and dependency relationships, and ledger entries would be stored.Yes.</blockquote>