On Sat, Mar 22, 2025 at 10:22:37PM -0500, Sami Imseih wrote:
> I think plan_node_id is probably the least controversial because that value
> comes straight from core, and different extensions cannot have their own
> interpretation of what that value could be.
Depends. An extension can plug in what they want. The point is that
the key used to identify a single plan is up to what extensions think
is relevant in a plan. That's heavily subject to interpretation.
What's not really subject to interpretation is that an extension
cannot know it should set and/or use as key identifier without
something that some portion pf the code structures knows about, or
these extensions have an inter-dependency. Anyway, there are also the
arguments about the set timing, reset timing, the extended protocol
argument, etc. So I've applied the patch for now, to start with
something.
> FWIW, Lukas did start a Wiki [0] to open the discussion for what parts
> of the plan should be used to compute a plan_id, and maybe we can
> in the future compite a plan_id in core by default.
Let's see where this leads.. I suspect that this is going to take
some time, assuming that we're ever able to settle on a clear
definition. Perhaps we will, or perhaps we will not.
--
Michael
