Schema Diff: ignore defseqrelid to avoid false SERIAL/BIGSERIAL diffs and invalid ALTER SQL (#10236)

[agu2347]

Bug

Schema Diff reports a false-positive difference (and generates invalid SQL) for structurally identical SERIAL/BIGSERIAL/SMALLSERIAL columns across two databases.

Fixes #10236

Root cause

get_formatted_columns() reprojects a column as SERIAL/BIGSERIAL/SMALLSERIAL when it owns (pg_depend, deptype='a' -> seqrelid) the same sequence referenced by its own nextval() DEFAULT (pg_attrdef -> defseqrelid; see #9896/#10100/#10101). Both seqrelid and defseqrelid survive into the column dict that Schema Diff compares, but they are raw sequence OIDs -- meaningless to compare across two independently-created databases. Two databases can have byte-for-byte identical BIGSERIAL columns while their owned sequences get assigned completely different OIDs.

SchemaDiffTableCompare.column_keys_to_ignore already excluded seqrelid, but not its sibling defseqrelid. That gap caused two symptoms from the same root cause:

1. compare() / are_dictionaries_identical() reports a table as different purely because of the benign defseqrelid mismatch, even when every user-visible column property (including cltype) is identical.
2. compare_target_cols() adds the column to the changed diff list. Downstream SQL generation (columns/sql/*/update.sql) then renders ALTER COLUMN ... TYPE {cltype} for that column -- even though data.cltype == o_data.cltype ('bigserial' on both sides) -- producing the invalid statement reported in the issue: ALTER COLUMN adl_id TYPE bigserial;.

Fix

Add 'defseqrelid' to SchemaDiffTableCompare.column_keys_to_ignore, alongside its existing sibling 'seqrelid'. This fixes both symptoms at the source: the column is no longer considered different at all, so neither the false-positive diff nor the invalid ALTER COLUMN statement is generated.

Testing

Added tests/test_schema_diff_serial_column_ignore.py, mirroring the existing columns/tests/test_serial_detection_unit.py unit-test convention (BaseTestGenerator + mock dict fixtures, no live DB needed for the logic under test):

• Two BIGSERIAL columns differing only by defseqrelid are not flagged as added/changed.
• A genuine difference (NOT NULL toggled) on an otherwise-identical BIGSERIAL column is still detected, proving the fix doesn't mask real diffs.

Since standing up the full Postgres-backed regression harness in my sandbox was impractical for this change, I additionally verified the actual patched SchemaDiffTableCompare.compare_target_cols() logic in isolation (loading the real class source directly, bypassing only the unrelated Flask/DB-heavy base-class import) with 6 checks, including a check that reverting the fix reproduces the exact false positive described in the issue (proving this is a genuine regression guard, not a tautological test). All passed.

pycodestyle --config=.pycodestyle (the project's actual CI style gate, confirmed via .github/workflows/check-python-style.yml) is clean on both changed files.

Summary by CodeRabbit

• Bug Fixes
  • Fixed schema comparisons for serial columns across independently created databases.
  • Prevented differing sequence identifiers from being reported as false differences or generating invalid alteration statements.
  • Genuine column changes, such as NOT NULL differences, continue to be detected.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: ef89a3b2-abcf-4716-aeb9-be147fc5f903

📥 Commits

Reviewing files that changed from the base of the PR and between 067f7af and 229177a.

📒 Files selected for processing (2)

• web/pgadmin/browser/server_groups/servers/databases/schemas/tables/schema_diff_table_utils.py
• web/pgadmin/browser/server_groups/servers/databases/schemas/tables/tests/test_schema_diff_serial_column_ignore.py

---

Walkthrough

The schema diff now ignores defseqrelid and existing seqrelid values when comparing serial columns. New BIGSERIAL tests verify that sequence OID differences are ignored while genuine column differences remain detected.

Changes

Serial schema-diff handling

Layer / File(s)	Summary
Ignore sequence OIDs in column comparison web/pgadmin/browser/server_groups/servers/databases/schemas/tables/schema_diff_table_utils.py, web/pgadmin/browser/server_groups/servers/databases/schemas/tables/tests/test_schema_diff_serial_column_ignore.py	The column comparison ignores defseqrelid alongside seqrelid. BIGSERIAL tests verify ignored sequence OID differences, genuine difference detection, and matched-column consumption.

Estimated code review effort: 2 (Simple) | ~10 minutes

Possibly related PRs

• pgadmin-org/pgadmin4#10167: Both changes handle SERIAL metadata. This PR updates schema-diff comparison, while the related PR addresses SERIAL detection and script generation.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly identifies the Schema Diff fix and the false SERIAL/BIGSERIAL differences and invalid SQL it addresses.
Linked Issues check	✅ Passed	The change ignores harmless sequence OID differences and adds tests that preserve detection of genuine column differences, meeting issue #10236.
Out of Scope Changes check	✅ Passed	The implementation and tests directly support the linked issue and PR objectives without unrelated changes.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}