Re: [PATCH] Rewrite undirected edge patterns in GRAPH_TABLE using UNION ALL

On Thu, Jul 30, 2026 at 9:20 PM Ayoub Kazar <ayoub(dot)kazar(at)data-bene(dot)io>
wrote:

>
> Also, given that the OR expansion was later replaced by cost based OR
> expansion, the transformation may not always yield optimal results. So, we
> should not always rewrite the undirected edge as UNION as the patch does.
>
> As mentioned above from what i noticed from some tests on Oracle query
> plans nearly everytime an OR expansion was applied, given both this patch's
> idea (UNION ALL only on edge table to generate two directions data -> then
> join separately) is a special case of the OR expansion, i find it
> interesting to see if we can try to prove that in all cases we would want
> this instead of the normal OR that mostly implies the bitmap scan ; which
> might suggest handling this "easier" case in rewrite for graph queries.
>

If OR transformation is implemented in the query optimizer, the
implementation in graph query is useless and, more seriously, misleading if
the correct statistics were to favour retaining OR.

--
Best Wishes,
Ashutosh Bapat