Schema Management

This guide covers managing your database schema beyond basic migrations, including Row Level Security (RLS), functions, triggers, and type safety.

Table Design Principles

Naming Conventions

Tables
Columns
Constraints

-- Use snake_case plural nouns
CREATE TABLE public.students (...);
CREATE TABLE public.student_school_interactions (...);
CREATE TABLE public.partner_entities (...);

-- Not: Student, StudentSchoolInteraction, partnerEntities

Standard Table Structure

Every table should include these common patterns:

CREATE TABLE public.your_table (
  -- Primary key
  id uuid PRIMARY KEY DEFAULT gen_random_uuid(),
  
  -- Your columns here
  name text NOT NULL,
  description text,
  
  -- Audit fields
  created_at timestamptz DEFAULT now(),
  updated_at timestamptz DEFAULT now(),
  
  -- Soft delete (optional)
  deleted_at timestamptz
);

-- Enable RLS
ALTER TABLE public.your_table ENABLE ROW LEVEL SECURITY;

-- Add updated_at trigger
CREATE TRIGGER update_your_table_updated_at
  BEFORE UPDATE ON public.your_table
  FOR EACH ROW
  EXECUTE FUNCTION update_updated_at_column();

-- Add indexes for foreign keys and common queries
CREATE INDEX idx_your_table_created_at ON public.your_table(created_at DESC);

Row Level Security (RLS)

RLS Strategy

RLS is your first line of defense. Every table should have RLS enabled with appropriate policies.

Common RLS Patterns

-- Anyone can read (careful with this!)
CREATE POLICY "Public read access"
  ON public.schools
  FOR SELECT
  USING (true);

-- Only non-deleted records
CREATE POLICY "Public read non-deleted"
  ON public.schools
  FOR SELECT
  USING (deleted_at IS NULL);

Testing RLS Policies

Always test your policies in the Supabase dashboard:

Navigate to Auth Policies

Go to Database → Tables → Select table → Policies

Use Policy Editor

Click “New policy” and use the template editor

Test with Different Users

Use “Test policy” with different user contexts

Check SQL Directly

-- Test as anonymous
SET request.jwt.claim.sub TO NULL;
SELECT * FROM public.your_table;

-- Test as specific user
SET request.jwt.claim.sub TO 'user-uuid-here';
SELECT * FROM public.your_table;

Database Functions

Utility Functions

-- Update timestamp trigger function
CREATE OR REPLACE FUNCTION update_updated_at_column()
RETURNS TRIGGER AS $$
BEGIN
  NEW.updated_at = now();
  RETURN NEW;
END;
$$ LANGUAGE plpgsql;

-- Soft delete function
CREATE OR REPLACE FUNCTION soft_delete()
RETURNS TRIGGER AS $$
BEGIN
  NEW.deleted_at = now();
  RETURN NEW;
END;
$$ LANGUAGE plpgsql;

-- Generate slug from name
CREATE OR REPLACE FUNCTION generate_slug(name text)
RETURNS text AS $$
BEGIN
  RETURN lower(
    regexp_replace(
      regexp_replace(name, '[^a-zA-Z0-9\s-]', '', 'g'),
      '\s+', '-', 'g'
    )
  );
END;
$$ LANGUAGE plpgsql;

Business Logic Functions

-- Calculate match percentage
CREATE OR REPLACE FUNCTION calculate_match_score(
  p_student_id uuid,
  p_school_id uuid
) RETURNS integer AS $$
DECLARE
  v_score integer := 0;
  v_student record;
  v_school record;
BEGIN
  -- Get student and school data
  SELECT * INTO v_student FROM public.students WHERE id = p_student_id;
  SELECT * INTO v_school FROM public.schools WHERE id = p_school_id;
  
  -- GPA match (25 points)
  IF v_student.gpa >= v_school.avg_gpa - 0.5 THEN
    v_score := v_score + 25;
  END IF;
  
  -- Size preference (25 points)
  IF v_student.preferred_size = 
    CASE 
      WHEN v_school.size < 5000 THEN 'Small'
      WHEN v_school.size < 15000 THEN 'Medium'
      ELSE 'Large'
    END 
  THEN
    v_score := v_score + 25;
  END IF;
  
  -- Add more matching logic...
  
  RETURN v_score;
END;
$$ LANGUAGE plpgsql;

-- Get conversation participants
CREATE OR REPLACE FUNCTION get_conversation_participants(
  p_conversation_id uuid
) RETURNS TABLE (
  participant_id uuid,
  participant_type text,
  participant_name text
) AS $$
BEGIN
  RETURN QUERY
  SELECT 
    COALESCE(cp.student_id, cp.partner_user_id) as participant_id,
    cp.role as participant_type,
    CASE 
      WHEN cp.student_id IS NOT NULL THEN s.name
      WHEN cp.partner_user_id IS NOT NULL THEN pu.full_name
    END as participant_name
  FROM public.conversation_participants cp
  LEFT JOIN public.students s ON cp.student_id = s.id
  LEFT JOIN public.partner_users pu ON cp.partner_user_id = pu.id
  WHERE cp.conversation_id = p_conversation_id;
END;
$$ LANGUAGE plpgsql;

Triggers

Common Trigger Patterns

-- Automatic updated_at
CREATE TRIGGER update_students_updated_at
  BEFORE UPDATE ON public.students
  FOR EACH ROW
  EXECUTE FUNCTION update_updated_at_column();

-- Cascade soft deletes
CREATE TRIGGER cascade_soft_delete_messages
  AFTER UPDATE OF deleted_at ON public.conversations
  FOR EACH ROW
  WHEN (NEW.deleted_at IS NOT NULL AND OLD.deleted_at IS NULL)
  EXECUTE FUNCTION soft_delete_related_messages();

-- Maintain denormalized counts
CREATE TRIGGER update_conversation_message_count
  AFTER INSERT OR DELETE ON public.messages
  FOR EACH ROW
  EXECUTE FUNCTION update_message_count();

-- Audit trail
CREATE TRIGGER audit_student_changes
  AFTER INSERT OR UPDATE OR DELETE ON public.students
  FOR EACH ROW
  EXECUTE FUNCTION create_audit_entry();

Function for Triggers

-- Soft delete related records
CREATE OR REPLACE FUNCTION soft_delete_related_messages()
RETURNS TRIGGER AS $$
BEGIN
  UPDATE public.messages
  SET deleted_at = NEW.deleted_at
  WHERE conversation_id = NEW.id
    AND deleted_at IS NULL;
  RETURN NEW;
END;
$$ LANGUAGE plpgsql;

-- Update denormalized count
CREATE OR REPLACE FUNCTION update_message_count()
RETURNS TRIGGER AS $$
BEGIN
  IF TG_OP = 'INSERT' THEN
    UPDATE public.conversations
    SET message_count = message_count + 1,
        last_message_at = NEW.created_at
    WHERE id = NEW.conversation_id;
  ELSIF TG_OP = 'DELETE' THEN
    UPDATE public.conversations
    SET message_count = message_count - 1
    WHERE id = OLD.conversation_id;
  END IF;
  RETURN NULL;
END;
$$ LANGUAGE plpgsql;

Type Safety with TypeScript

Generate Types from Database

# Generate TypeScript types
supabase gen types typescript --project-id your-project-ref > types/database.ts

Using Generated Types

// types/database.ts (generated)
export interface Database {
  public: {
    Tables: {
      students: {
        Row: {
          id: string
          name: string
          email: string | null
          created_at: string
        }
        Insert: {
          id?: string
          name: string
          email?: string | null
          created_at?: string
        }
        Update: {
          id?: string
          name?: string
          email?: string | null
          created_at?: string
        }
      }
    }
  }
}

// Using in your app
import { Database } from '@/types/database'

type Student = Database['public']['Tables']['students']['Row']
type NewStudent = Database['public']['Tables']['students']['Insert']

// With Supabase client
const { data: students } = await supabase
  .from('students')
  .select('*')
  .returns<Student[]>()

Views and Materialized Views

Creating Views for Complex Queries

-- Student dashboard view
CREATE OR REPLACE VIEW public.student_dashboard AS
SELECT 
  s.id,
  s.name,
  s.email,
  COUNT(DISTINCT ssi.school_id) as schools_viewed,
  COUNT(DISTINCT CASE WHEN ssi.liked = true THEN ssi.school_id END) as schools_liked,
  COUNT(DISTINCT c.id) as active_conversations,
  COUNT(DISTINCT sch.id) as scholarships_saved
FROM public.students s
LEFT JOIN public.student_school_interactions ssi ON s.id = ssi.student_id
LEFT JOIN public.conversations c ON s.id = c.initiator_id AND c.initiator_type = 'student'
LEFT JOIN public.student_scholarship_interactions sch ON s.id = sch.student_id AND sch.saved = true
GROUP BY s.id, s.name, s.email;

-- Grant appropriate access
GRANT SELECT ON public.student_dashboard TO authenticated;

Materialized Views for Performance

-- School statistics that update periodically
CREATE MATERIALIZED VIEW public.school_stats AS
SELECT 
  sch.id,
  sch.name,
  COUNT(DISTINCT ssi.student_id) as total_views,
  COUNT(DISTINCT CASE WHEN ssi.liked = true THEN ssi.student_id END) as total_likes,
  AVG(CASE WHEN ssi.match_score IS NOT NULL THEN ssi.match_score::numeric END) as avg_match_score,
  COUNT(DISTINCT m.id) as total_messages
FROM public.schools sch
LEFT JOIN public.student_school_interactions ssi ON sch.id = ssi.school_id
LEFT JOIN public.partner_entities pe ON sch.id = pe.school_id
LEFT JOIN public.conversations c ON pe.id = c.partner_entity_id
LEFT JOIN public.messages m ON c.id = m.conversation_id
GROUP BY sch.id, sch.name;

-- Create index for fast lookups
CREATE INDEX idx_school_stats_id ON public.school_stats(id);

-- Refresh periodically (in a scheduled function or cron job)
REFRESH MATERIALIZED VIEW CONCURRENTLY public.school_stats;

Performance Optimization

Indexing Strategy

Always Index Foreign Keys

CREATE INDEX idx_messages_conversation 
ON public.messages(conversation_id);

Index Filter Columns

CREATE INDEX idx_students_graduation_year 
ON public.students(graduation_year);

Composite Indexes

CREATE INDEX idx_interactions_student_school 
ON public.student_school_interactions(student_id, school_id);

Partial Indexes

CREATE INDEX idx_active_students 
ON public.students(created_at) 
WHERE deleted_at IS NULL;

Query Performance Tips

-- Use EXPLAIN ANALYZE to understand query plans
EXPLAIN ANALYZE
SELECT * FROM public.students
WHERE graduation_year = 2024
  AND gpa > 3.5;

-- Create covering indexes for common queries
CREATE INDEX idx_students_search 
ON public.students(graduation_year, gpa) 
INCLUDE (name, email);

-- Use proper JOIN order (smallest dataset first)
SELECT * FROM public.messages m
JOIN public.conversations c ON m.conversation_id = c.id  -- messages first (larger)
JOIN public.students s ON c.initiator_id = s.id         -- then students (smaller)
WHERE s.id = 'specific-id';

Troubleshooting Schema Issues

Table already exists errors

Always use IF NOT EXISTS:

CREATE TABLE IF NOT EXISTS public.table_name (...);

Cannot drop table due to dependencies

Use CASCADE carefully:

-- See what depends on the table
SELECT * FROM pg_depend WHERE refobjid = 'public.table_name'::regclass;

-- Drop with cascade (careful!)
DROP TABLE public.table_name CASCADE;

RLS blocking all access

Check if RLS is enabled but no policies exist:

-- Check RLS status
SELECT tablename, rowsecurity 
FROM pg_tables 
WHERE schemaname = 'public';

-- Temporarily disable for debugging
ALTER TABLE public.table_name DISABLE ROW LEVEL SECURITY;

Function not working as expected

Debug with RAISE NOTICE:

CREATE OR REPLACE FUNCTION debug_function()
RETURNS void AS $$
DECLARE
  v_test text;
BEGIN
  v_test := 'Hello';
  RAISE NOTICE 'Value of v_test: %', v_test;
END;
$$ LANGUAGE plpgsql;

Next, learn about data operations or troubleshooting common issues.

Getting Started

Developer Guide

Web App

iOS App

Matching Algorithm

Database

AI Development

Schema Management

Schema Management

Table Design Principles

Naming Conventions

Standard Table Structure

Row Level Security (RLS)

RLS Strategy

Common RLS Patterns

Testing RLS Policies

Database Functions

Utility Functions

Business Logic Functions

Triggers

Common Trigger Patterns

Function for Triggers

Type Safety with TypeScript

Generate Types from Database

Using Generated Types

Views and Materialized Views

Creating Views for Complex Queries

Materialized Views for Performance

Performance Optimization

Indexing Strategy

Always Index Foreign Keys

Index Filter Columns

Composite Indexes

Partial Indexes

Query Performance Tips

Troubleshooting Schema Issues

Getting Started

Developer Guide

Web App

iOS App

Matching Algorithm

Database

AI Development

​Schema Management

​Table Design Principles

​Naming Conventions

​Standard Table Structure

​Row Level Security (RLS)

​RLS Strategy

​Common RLS Patterns

​Testing RLS Policies

​Database Functions

​Utility Functions

​Business Logic Functions

​Triggers

​Common Trigger Patterns

​Function for Triggers

​Type Safety with TypeScript

​Generate Types from Database

​Using Generated Types

​Views and Materialized Views

​Creating Views for Complex Queries

​Materialized Views for Performance

​Performance Optimization

​Indexing Strategy

Always Index Foreign Keys

Index Filter Columns

Composite Indexes

Partial Indexes

​Query Performance Tips

​Troubleshooting Schema Issues

Schema Management

Table Design Principles

Naming Conventions

Standard Table Structure

Row Level Security (RLS)

RLS Strategy

Common RLS Patterns

Testing RLS Policies

Database Functions

Utility Functions

Business Logic Functions

Triggers

Common Trigger Patterns

Function for Triggers

Type Safety with TypeScript

Generate Types from Database

Using Generated Types

Views and Materialized Views

Creating Views for Complex Queries

Materialized Views for Performance

Performance Optimization

Indexing Strategy

Query Performance Tips

Troubleshooting Schema Issues