• More structured documentation
• More detailed scenarios to make easier to understand

Difficult to determine the relationship between versions and features.

It might be something like:

• Multiple Sidebars: vuejs/vitepress#109 (comment)
• Extension: vuejs/vitepress#109 (comment)

• Set [Build PostgreSQL image with pgvecto.rs] as optional
• Add recommend image name imageName: <your-postgresql-image-repo>:<pg-tag> to ghcr.io/tensorchord/cloudnative-pgvecto.rs:16-v0.2.0
• Add step to apply yaml: kubectl apply -f cnpg.yaml at [Create PostgreSQL cluster]

We need change 0.2.0 to 0.2.x to tell user our original upgrade procedure is still available.

More data type support (FP16/INT8/Sparse/Binary/Matroyoshka)

https://docs.pgvecto.rs/getting-started/installation.html Here we could mention https://github.com/tensorchord/pgvecto.rs/releases/download/v0.2.1/vectors-pg14_x86_64-unknown-linux-gnu_0.2.1.zip and other zip packages.

• fp16 example
• SQ example

• How to choose between HNSW and IVFFlat
• Typical scenarios and create index examples
• Quantization and fp16

Problem

When I follow the steps in develop.md, got this error.

ubuntu:~/pgvecto.rs$ cargo install cargo-pgrx@$(grep 'pgrx = {' Cargo.toml | cut -d '"' -f 2)
error: invalid package name: `https://github.com/tensorchord/pgrx.git`
    Use `cargo install --git https://github.com/tensorchord/pgrx.git` if you meant to install from a git repository.

Possible Causes

When I run this

ubuntu:~/pgvecto.rs$ grep 'pgrx = {' Cargo.toml
pgrx = { version = "0.11.3", default-features = false, features = [] }
pgrx = { git = "https://github.com/tensorchord/pgrx.git", branch = "v0.11.3-patch" }

Use the cat command to filter and get the following results

ubuntu:~/pgvecto.rs$ grep 'pgrx = {' Cargo.toml | cut -d '"' -f 2
0.11.3
https://github.com/tensorchord/pgrx.git

So I think it can be solved like this, just add 'version' string in grep command params.

ubuntu:~/pgvecto.rs$ grep 'pgrx = { version' Cargo.toml | cut -d '"' -f 2
0.11.3

If there is no deviation in my understanding, this PR can be merged

tensorchord/pgvecto.rs#255

We need to document:

• text representation of vector types
• memory representation of vector types
• binary representation of vector types
• subscript behaviors of vector types

Reference: https://github.com/percent4/embedding_rerank_retrieval