Robert Haas <robertmhaas@gmail.com> writes:
> I find both of your proposed solutions above to be pretty inelegant,
They are that. If we were working in a green field I'd not propose
such things ... but we aren't. I believe there are now a fair number
of out-of-core index AMs, so I'd rather not break all of them if we
don't have to.
> and I think if this problem occurred with a core AM, I'd push for an
> API break rather than accept the ugliness. "This path is not valid
> because the AM cannot support it", "this path is crazy expensive", and
> "the user told us not to do it this way" are three different things,
> and signalling two or more of them in the same way muddies the water
> in a way that I don't like.
I think it's not that bad, because we can limit the knowledge of this
hack to the amcostestimate interface, which doesn't really deal in
"the user told us not to do it this way" at all. That argues against
my first proposal though (having amcostestimate touch disabled_nodes
directly). I now think that a reasonable compromise is to say that
setting indexTotalCost to +Inf signals that "the AM cannot support
it". That's not conflated too much with the other case, since even a
crazy-expensive cost estimate surely ought to be finite. We can have
cost_index untangle that case into a separate failure return so that
the within-the-core-optimizer APIs remain clean.
While that would require hnsw to make a small code change (return
+Inf not DBL_MAX), that coding should work in back branches too,
so they don't even need a version check.
regards, tom lane
