From: zthhhhh (@zthhhhh) <noreply+zthhhhh@github.com>
To: pgvector/pgvector <noreply+pgvector-pgvector@github.com>
Subject: Re: [pgvector/pgvector] issue #761: Enhancement: multicolumn vector index design
Date: Tue, 04 Aug 2026 05:51:21 +0000
Message-ID: <gh-pgvector-pgvector-761-c5175143484@github.com> (raw)
In-Reply-To: <gh-pgvector-pgvector-761@github.com>
References: <gh-pgvector-pgvector-761@github.com>
> What technical or design blockers are preventing these approaches from moving forward,
Hi @anuj-kukunarapu . In my opinion, one challenge is PostgreSQL's current planner and executor interfaces do not provide a straightforward way to implement this entirely within pgvector.
For example, an ACORN-like implementation would first need to generate a bitmap index scan as subpath of HNSW IndexScan node.