fix: grant permission for pg_vector_index_stat in install sql

[usamoi]

• closes chore: Fix permission for pg_vector_index_stat  #412

[VoVAllen]

Please cherry pick this into 0.2 branch

[usamoi]

Users can do it by themselves if they meet issues.

…

________________________________ From: Jinjing Zhou ***@***.***> Sent: Monday, March 11, 2024 3:02:45 PM To: tensorchord/pgvecto.rs ***@***.***> Cc: usamoi ***@***.***>; State change ***@***.***> Subject: Re: [tensorchord/pgvecto.rs] fix: grant permission for `pg_vector_index_stat` in install sql (PR #416) Please cherry pick this into 0.2 branch — Reply to this email directly, view it on GitHub<#416 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AS426HUFVFC7XEVODAU4SATYXVJJLAVCNFSM6AAAAABEPQLDPGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSOBXG42DQMZYGY>. You are receiving this because you modified the open/close state.Message ID: ***@***.***>

[PrivatePuffin]

Please cherry pick this into 0.2 branch

This is not a full non-beta release product. It's pretty much hot-but-good code, when implementing hot-new versions, you should do a lot of testing to ensure bugs like these don't hit your codebase. That's the burden of using pre-release projects like Vecto.rs.
They cannot send-out patch releases for every single bugfix.

Best blame Immich, i'm betting that's the source of your troubles, for overzealusly implementing pgvecto.rs without thorough superuser-and-non-superuser testing.

[VoVAllen]

@Ornias1993 Yeah, The diversity of user deployment methods for immich is something we hadn't met before. And the technique behind context search with filter is based on a new paper VBASE published in 2023. That's also the reason why we looks kind of in the beta status.

Luckily (or a bit sadly), all the complains so far are about the upgrade/installation problems, but not the function itself. We'll work closely with immich team to improve the user experiences.