Discussion about this post

User's avatar
Qian Arthur Wang's avatar

I agree that layout changes are rarely worth the trouble. It really comes down to the migration tax. In my experience building systems, the riskiest move is chasing theoretical peak performance while ignoring the cognitive overhead of breaking a standard protocol. Splitting and tenting are pragmatic in-place upgrades. They fix the physical bottleneck without a breaking API change to muscle memory. I deal with enough breaking changes at work as it is.

Ethan Tecson's avatar

Looking forward to it!

5 more comments...

No posts

Ready for more?