PostgreSQL Schema Display: Should schema names be included in table names?

[MH4GF]

Background

We received feedback from tokuhirom-san on Twitter suggesting that our current approach to handling PostgreSQL schemas might not be optimal for PostgreSQL users. The feedback indicates that PostgreSQL schemas should be displayed as part of the table names (similar to how tbls handles it), rather than being stripped out during parsing.

Original feedback: https://x.com/tokuhirom/status/1941857822784029145

Reference: tbls approach to PostgreSQL schemas

Current Behavior

Currently, Liam ERD strips schema qualifiers from PostgreSQL table names during parsing:

• A table named auth.users is displayed as just users
• A table named analytics.page_views is displayed as just page_views
• Schema context is completely lost in the visualization

Proposed Improvement

Display PostgreSQL table names with their schema prefix, similar to how tbls handles it:

• auth.users would be displayed as auth.users
• analytics.page_views would be displayed as analytics.page_views
• public.orders would be displayed as public.orders (or potentially just orders if it's the default schema)

Technical Analysis

From our codebase analysis:

Current Implementation:

• PostgreSQL parser strips schema qualifiers in converter.ts
• Table names are stored without schema prefixes
• ERD visualization shows unqualified table names
• Foreign key relationships work but lose schema context

Available Data Sources:

• TBLS format includes schema-qualified names: "public.users", "administrator.blogs"
• SQL dump files contain proper schema information
• PostgreSQL AST has schema data but we currently ignore it

Discussion Points

1. User Experience: Would PostgreSQL users prefer seeing schema-qualified table names?
2. Consistency: Should this be PostgreSQL-specific behavior or apply to all database types that support schemas?
3. Configuration: Should this be configurable (show/hide schema prefixes)?
4. Default Schema: Should we show the schema prefix for the default public schema?
5. Visual Design: How should schema-qualified names be displayed in the ERD? (e.g., schema.table, highlighting, grouping)

Potential Benefits

• Clarity: Users can immediately see which schema each table belongs to
• Disambiguation: Tables with the same name in different schemas won't be confused
• PostgreSQL Native Feel: More familiar to PostgreSQL users who are used to schema-qualified references
• Better Context: Easier to understand the logical organization of the database

Potential Drawbacks

• Longer Names: Schema-qualified names might make the ERD more cluttered
• Breaking Change: Existing users might be surprised by the change
• Database Agnostic: Other database systems might not have the same schema concept

Questions for Discussion

1. Should we implement this change for PostgreSQL specifically?
2. How should we handle the default public schema - show it or hide it?
3. Should this be a user-configurable option?
4. Any other considerations for PostgreSQL schema handling?

Related Work

• tbls PostgreSQL schema handling
• Current implementation in frontend/packages/db-structure/src/parser/sql/postgresql/

Looking forward to hearing thoughts from the community, especially from PostgreSQL users who can share their preferences and use cases.

[MH4GF]

📝

• schemaspy generates a page for each schema:
  • https://mastodon-schemaspy-rh9gi6jmg-route-06-core.vercel.app/index.html

[MH4GF]

• It should still be omitted in nodes, as it causes problems with the width of table nodes
  • public.user_...., only the common schema name is always visible, and table names are easily omitted.
• Where TableDetail shows users.id as public.users.id, that's fine.
• Which use case is more prevalent: one that uses only one schema, such as public, or one that actively utilizes the schema?