Skip to content
DevMeme
1423 of 7590
Performance Post #1596 · source on Telegram

The Duality of Developer Logic on Bundle Size and Caching

Description

A screenshot of two consecutive tweets from user Harry Roberts (@csswizardry). Both tweets highlight contradictory statements from developers about web performance. The first tweet says: 'Developers: We don't care about bundle size - users download it once and then it's cached. Developers: We release 10 times a day.' The second tweet continues the pattern: 'Developers: We don't care about bundle size - users download it once and then it's cached. Developers: I haven't written a cache-control header in my life.' The humor comes from the blatant hypocrisy and logical fallacies presented. Senior developers find this relatable and funny because it exposes common, flawed justifications for ignoring performance best practices. Frequent releases invalidate caches, and without proper cache-control headers, caching is ineffective, making the initial argument completely moot

Comments

7
Anonymous ★ Top Pick Some developers treat caching like a religion: they believe in its power to absolve them of their sins (large bundles), but they've never actually read the holy book (the RFCs for Cache-Control)
  1. Anonymous ★ Top Pick

    Some developers treat caching like a religion: they believe in its power to absolve them of their sins (large bundles), but they've never actually read the holy book (the RFCs for Cache-Control)

  2. Anonymous

    Roadmap update: “Bundles are cached forever, so ship at will.” Fourteen deploys later, our missing Cache-Control header has turned every 3 MB app.js into an ad-hoc push notification system over 4G

  3. Anonymous

    The real cache invalidation problem isn't naming things - it's explaining to stakeholders why their 'just ship it fast' mentality means users are downloading your entire React app every lunch break because you deploy faster than their browser can say '304 Not Modified'

  4. Anonymous

    Ah yes, the classic developer trifecta: assume aggressive browser caching will save users, deploy 10 times daily invalidating said cache, and never actually configure cache-control headers. It's like claiming your microservices are stateless while running a monolithic database with a single point of failure - technically possible, just profoundly missing the point. The real kicker? We've all been this developer at some point, confidently explaining to stakeholders how our 5MB bundle 'only downloads once' while our CI/CD pipeline gleefully busts that cache every hour

  5. Anonymous

    Continuous deployment without immutable assets is just expensive distributed cache invalidation therapy

  6. Anonymous

    CI/CD: Continuous Integration/Constant Download - “bundle size is fine, it’s cached,” says the team deploying hourly with zero Cache-Control

  7. Anonymous

    Cache invalidation and naming things are hard; frequent deploys without Cache-Control? That's the senior dev rite of passage

Use J and K for navigation