@shiok If you want you can run maintainers/scripts/haskell/regenerate-haskell-packages.sh to immediately see and test the consequence of your change. (And then you can also put that into the commit.)

This builds fine for me. If you sqash this I can merge it. I would love if you could run the above mentioned script from the root of your nixpkgs repo as part of your commit. But I can run it afterwards if you want.

@maralorn Thanks; I have squashed the commits now. Also included is the result of regenerate-hackage-packages.sh (although FYI, I had to do export LC_ALL=C.UTF-8 to avoid a certain commitBuffer: invalid argument (invalid character) error on my Debian 10).

@shlok Thank you for testing the script!

Huh, that seems worth investigating. Do you think it would be smart to set that variable in the beginning of the script to prevent this from happening for other users?

Thanks for the PR!

@shlok Thank you for testing the script!

My pleasure :)

Do you think it would be smart to set that variable in the beginning of the script to prevent this from happening for other users?

I'm not sure about that. For me it was a mere workaround based on an internet search that led me to commercialhaskell/stack#793

Shows to double check, if you can tell: Does the fail happen while creating the nix-shell the script starts or this an error that hackage2nix throws?

It's thrown by hackage2nix:

hackage2nix: /home/user/nixpkgs/pkgs/development/haskell-modules/hackage-packages.nix: commitBuffer: invalid argument (invalid character)

(I added export LC_ALL=C.UTF-8 just before the hackage2nix line to "fix" the issue.)

By the way, I just noticed that LC_ALL is not set at all before my export line. It however is set on my local machine, so I guess this is a nix-shell thing.

I imagine that this LC_ALL=C.UTF-8 issue is the same as #64603 (comment).

We could probably set LC_ALL=C.UTF-8 in regenerate-hackage-packages.sh for users that don't have this already set on their system.

I tried adressing this in 66de11d (#122117)