Re: Failure while inserting parent tuple to B-tree is not fun

On Thu, Jan 23, 2014 at 1:36 PM, Peter Geoghegan <pg(at)heroku(dot)com> wrote:
> So while post-recovery callbacks no longer exist for any
> rmgr-managed-resource, 100% of remaining startup and cleanup callbacks
> concern the simple management of memory of AM-specific recovery
> contexts (for GiST, GiN and SP-GiST). I have to wonder if there isn't
> a better abstraction than that, such as a generic recovery memory
> context, allowing you to retire all 3 callbacks. I mean, StartupXLOG()
> just calls those callbacks for each resource at exactly the same time
> anyway, just as it initializes resource managers in precisely the same
> manner earlier on. Plus if you look at what those AM-local memory
> management routines do, it all seems very simple.

What are your thoughts on this, as someone that has a broader
perspective here? Are you inclined to keep the startup and cleanup
callbacks in anticipation of a day when that degree of generality is
useful? That would be pretty well-precedented of course, but I would
like to hear your opinion.

--
Peter Geoghegan