On Mon, Sep 6, 2021 at 7:09 PM Thomas Munro <thomas.munro@gmail.com> wrote:
> FTR I have a patch in development that adds a demand-paged (though
> Anastasia recently suggested reconsidering that) per-SMGR relation
> shmem object pool that initially tracks precisely "how long it is",
> but I hope it would provide a hook to hang many other things on in
> future that need coordination at the relation level, such as
> synchronized scan position, and I hope, radix-based buffer mappings.
Cool!
> Could be relevant? https://commitfest.postgresql.org/34/2933/ (I'll
> rebase it soon).
I think that it is relevant. You're coming at the same thing from
roughly the opposite direction, at least in my mind. But that seems
like it might be a very good thing, given the specifics. I think that
we can meet each other in the middle, without there being any
particular risk of overlapping work or conflicting requirements. The
resource management aspects of my own prototype are the least worked
out and shakiest parts overall, so this seems likely. You've probably
already solved numerous problems for me.
--
Peter Geoghegan