A presentation at PragueJS in in Prague, Czechia by Robin Pokorny
2025, 30Oct = 24Dec I LOVED SNAPSHOTS IN 2016 A Decade Later, Let’s Revisit ROBIN POKORNY
TONIGHT 01 02 03 WHAT IS SNAPSHOT TESTING WHAT WENT WRONG WHAT NOW?
01 WHAT IS SNAPSHOT TESTING
02 WHAT WENT WRONG
‘The most common drawback is the fragility of snapshot tests’ —STIP PAPER
LACK OF DEVELOPER INTENTION Tests don’t encode what’s important GENERATED FILE PROBLEM Need manual verification LARGE SNAPSHOTS Developers don’t scrutinize them HIGH FALSE NEGATIVES Tests fail when code is actually fine
03 WHAT NOW?
USE ● ● ● ● ● API response structures (not UI) Error messages and CLI output for developer-facing tools Code transformations Small, stable components in design systems Legacy refactoring AVOID ● ● ● ● ● Large component trees Components with frequent changes “It renders” smoke tests (use linting instead) Testing behavior As primary testing strategy
https://selfie.dev
TAKEAWAY SNAPSHOTS PROVIDE VALUE WHEN USED EFFECTIVELY
WEB me@robinpokorny.com NAME IN/IG/YT Let’s connect
STATE OF JS https://survey.devographics.com/survey/state-of-js/2025
In 2016, snapshot testing felt revolutionary. I promoted it everywhere. Then came the backlash: fragile tests, large snapshots, and mindless change approval. The community wasn’t amazed. Yet in 2024, Node.js added native snapshot support. Why?
This talk traces snapshot testing’s turbulent decade: from Facebook’s game-changing release to general rejection, and examines what we got right, what we got wrong, and what place snapshots have today. Spoiler: the story isn’t as simple as “snapshots bad”. Let’s revisit with clearer eyes.
for free. You
can too.