| 18:24 | <shu> | bakkot: yes, we made the decision back then that they're capture-by-value |
| 18:24 | <shu> | since aliases, are, well, aliases, and not bindings |
| 18:25 | <shu> | if we come up with storage for algorithm steps, we could, but the work around is to allocate records that you use as boxes |