start command won't recover from interrupted initialization

[cbzehner]

Bug report

Describe the bug

When supabase init is interrupted or otherwise unavailable, the supabase start command isn't able to recover and provides a misleading error message.

To Reproduce

I'm on a 2019 MBP running Big Sur...

1. Turn off Docker globally (you may also need to clear any local Docker images used in Supabase's compose file)
2. Run supabase init taking the defaults and you'll end up with "Error running docker-compose."

❯ supabase init
✔ Port for Supabase URL: · 8000
✔ Port for PostgreSQL database: · 5432
✖ Error running docker-compose.

3. Start the Docker daemon again...
4. Try running supabase init (again). No-op.

❯ supabase init
Project already initialized. Remove .supabase to reinitialize.

5. Run supabase start and see the same error from the init command.

❯ supabase start
✖ Error running docker-compose.

This was solved by rm -rf .supabase and re-running supabase init but it isn't clear the underlying cause.

Expected behavior

Ideally supabase start would detect initialization had failed and prompt me to re-initialize or supabase init should detect a badly configured .supabase/ directory.

System information

• OS: macOS
• Version of supabase-cli: 0.3.6
• Version of Node.js: v15.11.0

[soedirgo]

Thanks for the report! I'll add proper cleanup for aborted init soon.

[github-actions]

🎉 This issue has been resolved in version 0.3.8 🎉

The release is available on:

• npm package (@latest dist-tag)
• GitHub release

Your semantic-release bot 📦🚀