There seem to be a lot of independent work items which are more difficult because of the current vmap / xenheap / domheap division; e.g., Jürgen’s domain memory pool series, ASI, etc. It seems like it would be good to sketch out a plan for where we want to reach eventually, and perhaps intermediate states which may be more helpful or less helpful for reaching that state.
Happy to consider expanding this to other topics that may be of interest to the same group of people.