Part 6 — History in Action

M
Written By Melih SünbülLast updated 5 months ago

Your app already has full bitemporal history — you didn't write a single line for it. Powered by django-simple-history under the hood, the framework records every change with two time dimensions: when it happened (valid time) and when we learned about it (system time). In this final part, you'll see it in action.

Scenario: Correcting an Expense After Quarter Close

It's February 5th. Anna from the Design team notices that her "Flight to Munich" expense was entered as €450, but the actual receipt shows €380. The quarter has already closed, but the correction needs to be recorded properly.

Edit the Expense

  1. Navigate to Expense in the frontend
  2. Find Anna's "Flight to Munich" record
  3. Change the amount from €450.00 to €380.00
  4. Save

Open the History Panel

Click the History icon on the record. You'll see:

Version

Amount

Valid From

Valid To

Changed By

v2 (current)

€380.00

Feb 5, 10:15 AM

—

Anna Schmidt

v1 (superseded)

€450.00

Jan 20, 2:30 PM

Feb 5, 10:15 AM

Anna Schmidt

Both values are preserved. The original €450 entry is never deleted — it's superseded. This is exactly what auditors need: a complete, tamper-proof trail.

Timeline Editing: Backdating the Correction

Now suppose Anna realizes the correct amount (€380) was actually valid since January 20 (when the flight happened), not since today. She needs to backdate the correction.

Edit valid_from in the History Panel

  1. In the history panel, click on v2 (the €380 version)
  2. Change valid_from from Feb 5 to Jan 20, 2:30 PM
  3. Save

The system recalculates the timeline:

Before:

Jan 20                              Feb 5
├── v1: €450.00 ────────────────────┤── v2: €380.00 ──────────►

After backdating:

Jan 20
├── v2: €380.00 ──────────────────────────────────────────────►

The v1 entry still exists in the system history (Level 2) — you can always answer: "On February 4th, what did the system think this expense was?" Answer: €450, because the backdated correction hadn't been entered yet.

What the Two Levels Show

Level 1 — Valid Time ("What was true?")

After the backdating, valid time shows:

Amount

Valid From

Valid To

€380.00

Jan 20, 2:30 PM

current

This is the business reality: the flight always cost €380.

Level 2 — System Time ("When did we know?")

System time tells the complete knowledge story:

Knowledge

Sys From

Sys To

"We thought it was €450"

Jan 20, 2:30 PM

Feb 5, 10:15 AM

"We now know it's €380"

Feb 5, 10:15 AM

current

You never lose information. Even after the backdating correction, both the original belief AND the correction are preserved in the system history.

Recalculate After the Correction

After correcting the expense:

  1. Navigate to Reports → BudgetSummary → Design team, Q1 2026
  2. Click Calculate ▶️

The calculation will reflect the corrected amount (€380 instead of €450). The calculation log shows the updated numbers — and the previous calculation log is still accessible in the history.

You're Done!

Congratulations! You've built a complete business application following the ETL pattern:

Layer

What You Built

Extract (Upload/)

Three upload models for CSV ingestion

Transform (Input/)

Three input models with validation and permissions

Load (Reports/)

Budget calculations with rich logging and dashboards

And here's how much code that took:

Feature

What You Wrote

3 input models

Three files in Input/

3 upload models

Three files in Upload/

1 serializer

One serializers.py for API validation

Automatic CRUD + API

Zero lines — powered by Django REST Framework

Budget calculations

One calculate() method

Rich logging

A few LexLogger calls

Validation

One pre_validation() method

Role-based permissions

Two permission_*() methods

Interactive dashboards

Two Streamlit methods

Full bitemporal history

Zero lines — powered by django-simple-history

Frontend data grid

Zero lines — powered by AG Grid

Timeline editing

Zero lines — built into the frontend

Total code: ~400 lines (including whitespace, docstrings, and imports). Everything else — the web UI, API, authentication, history, real-time updates — is provided by the framework. Lex App is open source — we recommend reading the source for deeper understanding.

What's Next?

Now that you've completed the tutorial, explore the rest of the documentation:

Lex App is open source. We recommend browsing the source code for deeper understanding of how the framework works under the hood.

Was this helpful?

Your feedback shapes what we write next.