Summary

Historic races (approx. 2018–2019) return Stints from the timing app API as a plain list instead of a dict with string indices. The previous parser assumed dict format unconditionally, causing all stints to be silently dropped when a list was received, no error, no warning.

Fixes #863, related to #860.

Root cause

In timing_app_data (fastf1/_api.py), the loop over stints used enumerate(update) and only reassigned stint_number and stint inside an isinstance(update, dict) check. When update was a list, the check was never entered, but the list itself was never normalized, causing downstream code to receive incomplete data.

Fix

Normalize the list format to dict before the loop:

if isinstance(update, list):
    update = {str(i): s for i, s in enumerate(update)}
for stint_number, stint in update.items():
    stint_number = int(stint_number)

This removes the ambiguous isinstance check inside the loop body and makes the two formats converge before any processing happens.

Validation

Tested against:

• Baku 2018 (legacy list format, lap-1 pit stops): all 5 affected drivers now show correct stint history
• Baku 2025 (modern dict format, lap-1 pit stops): no regression
• Monaco 2024 (modern dict format, red flag lap 1): no regression

Tests

Two unit tests added to fastf1/tests/test_api.py:

• test_timing_app_data_legacy_list_format: verifies list format is parsed correctly and no stints are dropped
• test_timing_app_data_modern_dict_format: verifies dict format continues to work as before