Post

> The first bad revision is: xkwomrzz a8f99ef1f6e550 go/printer: use architecture-independent math for alignment decisions As expected, you can blame floating point numbers for being terrible: go.dev/cl/752862 You are also today's lucky winner! This CL didn't change a single line in any Go repo.

go.devgo.devGerrit Code Review

0 reposts2 likes

To be clear, I don’t think this is a regression. Some calculations were adjusted slightly to ensure they are arch independent. You found an edge case where this actually resulted in a visible change. My guess is that on 1.26 this code needed to be one space wider to trigger the same behavior.