dbx-regenerate

DBX Regenerate

You are helping regenerate the DBX-generated code after changes to .dbx schema files.

Overview

DBX (Database Extension) is Storj's schema-first ORM that generates Go code from .dbx schema files. When developers modify .dbx files in satellite/satellitedb/dbx/, the generated code in satellitedb.dbx.go must be regenerated.

This skill:

1. Runs the DBX code generator
2. Shows a summary of what changed
3. Validates the generated code compiles
4. Reports any errors with helpful context

Instructions

Follow these steps to regenerate DBX code:

1. Check for Modified .dbx Files

First, check which .dbx files have been modified (optional but helpful for context):

git status satellite/satellitedb/dbx/

This helps understand what schema changes triggered the regeneration.

2. Run DBX Code Generation

Execute the code generation:

cd satellite/satellitedb/dbx && go generate

This runs the DBX compiler which:

• Parses all .dbx schema files
• Generates SQL for PostgreSQL, CockroachDB, and Spanner
• Produces Go code with type-safe methods
• Creates the ~50,000 line satellitedb.dbx.go file

Expected output: Should see messages about code generation progress.

3. Show Diff Summary

After generation completes, show a summary of what changed:

git diff --stat satellite/satellitedb/dbx/satellitedb.dbx.go

Also show a preview of the changes:

git diff satellite/satellitedb/dbx/satellitedb.dbx.go | head -100

What to look for:

• New methods added (e.g., Create_, Get_, Update_, Delete_ methods)
• Modified method signatures
• New model structs
• Changes to existing queries

4. Validate Compilation

Ensure the generated code compiles successfully:

go build ./satellite/satellitedb/dbx

If compilation succeeds, the regeneration was successful.

5. Report Results

Provide a clear summary to the user:

If successful:

• Confirm DBX regeneration completed
• Summarize what changed (e.g., "Added 3 new methods for the users table")
• Show key new/modified methods
• Confirm compilation succeeded
• Suggest next steps (e.g., "You may want to create a database migration for the schema changes")

If errors occurred:

• Show the full error output
• Identify which .dbx file has the issue (look for file references in error)
• Explain the error in plain language
• Suggest fixes based on common DBX errors (see below)

Common DBX Errors and Fixes

Syntax Error in .dbx File

Error pattern: parse error, unexpected token, invalid syntax

Cause: Incorrect .dbx syntax (missing commas, invalid field types, etc.)

Fix:

• Check the .dbx file line number mentioned in the error
• Verify syntax matches DBX documentation patterns
• Common issues: missing commas between fields, invalid field type names, incorrect query syntax

Duplicate Model/Query Definition

Error pattern: already defined, duplicate definition

Cause: Model or query defined multiple times across .dbx files

Fix:

• Search for the duplicate definition across all .dbx files
• Remove or rename the duplicate
• Ensure model names are unique across all .dbx files

Invalid Field Type

Error pattern: unknown type, invalid type

Cause: Using a field type that DBX doesn't recognize

Valid DBX types:

• text, blob, int, int64, uint, uint64
• bool, timestamp, float64
• utimestamp (microsecond timestamp)

Fix: Change the field type to a valid DBX type

Missing Primary Key

Error pattern: no primary key defined

Cause: Table definition missing a primary key

Fix: Add a primary key using ( key <field> ) in the model definition

Invalid Query Pattern

Error pattern: invalid query, unknown read pattern

Cause: Using an incorrect query pattern name

Valid patterns:

• read one, read all, read first, read paged, read limitoffset, read scalar
• update, delete, create, count

Fix: Use a valid query pattern from the list above

Circular Dependency

Error pattern: circular dependency, import cycle

Cause: .dbx files referencing each other in a circular way

Fix: Restructure the models to remove circular references

Post-Regeneration Checklist

After successful regeneration, remind the user to:

1. Review the diff: Check that generated changes match expectations
2. Create migration: If schema changed, create database migration files
3. Run linter: Ensure generated code passes linting

   make llint LINT_TARGET=./satellite/satellitedb/dbx
   

4. Commit changes: Commit both .dbx changes and generated code together

Understanding the Generated Code

The satellitedb.dbx.go file contains:

• Model structs: Go representations of database tables
• Field types: Type-safe field constructors (e.g., User_Id_Field)
• CRUD methods: Auto-generated database operations
• Backend implementations: Separate implementations for PostgreSQL, CockroachDB, and Spanner
• Transaction support: WithTx() methods for transactions
• Error handling: Wrapped errors with proper error codes

Example Workflow

User modified: satellite/satellitedb/dbx/user.dbx - added last_login timestamp field

Regeneration output:

DBX regenerated successfully!

Changes:
- Modified: User model (added field: last_login)
- New methods: Update_User_LastLogin_By_Id
- Lines changed: +156 -12

Generated methods:
- User.LastLogin field added
- User_LastLogin_Field type added
- Update methods now include last_login in optional fields

Compilation: ✓ Success

Next steps:
1. Review the diff to ensure changes are correct
2. Create a migration to add the last_login column
3. Run tests to ensure compatibility

Notes

• DBX generation can take 10-30 seconds depending on system performance
• The generated file is ~1.9 MB and ~50,000 lines of code
• Always regenerate after modifying ANY .dbx file
• The generator is located in satellite/satellitedb/dbx/gen/main.go
• DBX supports PostgreSQL, CockroachDB, and Spanner from the same schema
• Generated code includes monkit instrumentation for metrics
• All errors are wrapped with errs.Class("satellitedb")

Tips

• Use git diff to see exactly what changed before committing
• If regeneration seems slow, check for syntax errors first
• Keep .dbx files focused - one file per logical domain (users, projects, etc.)
• Test generated code with all supported backends if possible
• DBX errors are usually specific - read them carefully for the fix