Thread by @ChShersh

I'm so pissed by the situation around removing (/=) from the Eq typeclass in #Haskell 😡

👇🧵 A very angry thread and I why think that people who want to introduce similar breaking changes to Haskell are just a bunch of kids who want to play with their fancy toys 🎮🎳

DISCLAIMER: I'm not against breaking changes entirely. Sometimes they are unavoidable and there's no better to fix the problem.

For instance, the recent security vulnerability in `aeson`. The problem already affected ALL Haskell applications and needed to be fixed ASAP.

However, for almost all breaking changes in GHC or base there's no urgency at all. Breaking changes were introduced either if someone wanted to change things or if the new fancy feature is not compatible with the existing behaviour (I'm looking at you ImpredicativeTypes😒)

⚠️ !!! VERY IMPORTANT !!! ⚠️

By introducing unimportant breaking changes to GHC or core libraries, you ask volunteers to do MORE free work in their free time to maintain compatibility with unnecessary changes.

Haskell maintainers already sacrificed tons of their time to produce OSS libraries. But now you're telling them that they need to do even more work if they want their solutions to continue solving problems??

👆 When you break backwards compatibility, you completely disrespect other people, their time, book authors, existing tutorials, guides, newcomers.

Like, how a single minor non-fixing-anything change could be more important than ALL TUTORIALS AND OSS LIBRATIRES WRITTEN BEFORE???

You could say that things break in Haskell all the time so no big deal. But, okay, do newcomers aware of this fact? Do they enjoy working on a solution for a year, only to realize later that it won't work anymore with the new compiler for some reason?

No.

And now users of such a library must choose:

☹️ Stay with the old compiler forever
☹️ Maintain a fork of this library forever
☹️ Drop the dependency entirely

There's no win. It's simply not sustainable to ask volunteers to do even more work, even after many years!

What's even worse: there's no roadmap of changes! Or any plan where GHC is moving. Things can become broken in an incompatible way at ANY MOMENT by ANY PERSON. And it's purely random (but at least it's pure!).

Imagine considering such a risk when deciding to choose Haskell for your solution?

"Well, actually, the entire ecosystem can fall apart any time for no apparent benefit"