Skip to content

Meeting 2024‐03‐22

Alyx edited this page Mar 26, 2024 · 7 revisions

Workday sponsorship

They are sponsoring us to work on composite foreign keys. Waiting for confirmation before we start working on it.

Extraction of dialects

  • Do we want to move dialect-specific data types to dialect packages
    • If the DataType is supported by more than one dialect, it can go in core, otherwise it goes in dialect packages
  • Dialects could ensure that the core version matches the dialect version (so ideally they should be exactly equal)
  • Splitting tests is something we can look into later
  • Dialect toolkit
    • Basically a public API that exposes everything that's needed to create a new dialect for third-parties
    • Won't be released officially in v7. We'll work on it slowly but not a high priority.
  • We want to move iBMI and Snowflake to a community org because we have no way to integration test yet
    • Note: Rik managed to get a connection to an iBMI machine & log in. Worth looking into
    • Sadly we can't run more than smoke tests, as the machine is a community one and is very slow
    • Tried to reach out to snowflake but no answer

Oracle

  • Once we are ready to release v7, we need to reach out to the oracle team to update their v6 implementation
    • They were pausing development while working on their updated connector
    • Or maybe once the dialects are split, depends on our release strategy for v7

Cockroach

  • PR is pretty much done. It was split in two PRs (one without transactions and one that adds transactions).
  • Next step: Review it
  • They expressed interest in sponsoring us
  • They had a lot of merge conflicts due to how many major refactors we've done, such as moving to prettier, extracting utils to their own folders, and likely soon, extracting dialects

Yugabyte

  • Postponed until we do major improvements to our test suite run times

Latest v7 release plan

  • Two blockers remain: CLI (in progress) and extracting dialects (in progress)
    • After the dialect extraction, we will go into beta
    • We would like to merge @lohart13's migrating-to-typescript PRs before the stable release
  • We will continue to support v6 for 6 months after releasing v7
  • We are starting the CLI from scratch using oclif, TypeScript-first, ESM, etc
  • sequelize-auto will also be moved inside of the CLI
  • We want to add migration generation to the CLI (one that compares local models to the database tables and generates the SQL to update it)
  • For future major changes, what we want to do is:
    • Implement anything that requires structural changes now, such as extract dialects
    • Improvements to existing features that require breaking changes will be done by copying the method into one prefixed with UNSTABLE_. Once it's time to do a new major release (ideally once a year, in may-june), remove the prefix, and move the old implementation to the LEGACY_ prefix (for one major release), so people have an easier time updating, and we're not stuck working on v8 for 2 years and we can continue working on them in the v7 release
  • Do we want to do https://github.com/sequelize/sequelize/pull/17124 in v7?
    • Not a blocker so if it's ready before the stable release, good. If it's not, it will use the UNSTABLE_ prefix
  • Aiming for june for the beta release, but we had the same conversation two years ago. It's very difficult to plan a project you do on your free time

Misc Issues

Issues we talked about during this meeting (notes in issues)