One strength of go is very fast compiler, if it was trying to find if 6 levels down a call chain the map you pass is being assigned to it would be slower (as a nil map can still be read from)

This being said a m=set(m,k,v) working like append for slice would have been nice

It’s pretty much low quality bait and exaggeration. You won’t get any specifics.

What error do you think the compiler should catch and doesn’t for instance?

ic, 1 internal compiler error in 2021 and 1 this year, how is that once a month?

Can you link the last 2 or 3 crash you had and reported? I used go for 8 years and never had a crash, nor an unclear message.

Any specifics? govulncheck doesn't replace getting bug fixes and release notes nicely gathered by dependabot

I wish I could set org wide settings (like groups) instead of copy-pasta'ing the config in each github repo though

that tool does need a fixed Dockerfile and getting rid of stutter inducing cmd/ though, I’ll give you that (for docker: laurentsv.com/blog/2024/06... )

:) so what would you refactor in that package that you can’t? to me it seems a theoretical reason to use internal. It’s not to say that putting thoughts in API surfaces isn’t a good idea, just that I don’t think using internal/ “initially” is useful.

Had to sign up to disagree, most modules do not need internal/, it’s premature hiding and slows things down. Do you have an example of such package that can’t be changed?

More details/nuance about go layout: laurentsv.com/blog/2024/10...