Use the README to explain what this package is about

[DamienCassou]

The current README says how to use flake-compat but not what it is. It would also be nice to write a description in the project settings.

[shajra]

Adding to this, I only recently realized there has been a new builtins.getFlake for some time. I'm learning about these things by mimicking other projects. But I think that leads to substandard ideas sticking around.

It would be nice for the README to make clear whether builtins.getFlake is the preferred way to deal with flakes compatibility moving forward. If that's not the case, it would be good to discuss some pros/cons of each approach.

I think the only reason not to use builtins.getFlake is in case someone has a version of Nix older than 2.4. Anything else?

[shajra]

Coming back to this, I've realized that builtins.getFlake is only usable with flakes enabled. If a user is enabling the feature, I think they are just getting pulled into the world of flakes, properly. This project still seems to server a purpose. It would be good if the README couple explain this all better. I'm still figuring this out myself.

[eclairevoyant]

quite simply: this is a flakes polyfill written in pure nix instead of c++
and it's more or less a dead project

[link2xt]

I have found that ghostty is using it:
https://github.com/ghostty-org/ghostty/blob/aa2f31179be2ac8376581466d158f4a4f3df7996/default.nix

But could not understand what it does from the readme.

[eclairevoyant]

this is a flakes polyfill

[link2xt]

@eclairevoyant I got it, but this tells almost as much as "flakes-compat" name.

This line does a better job than readme at describing what it is:
https://github.com/NixOS/flake-compat/blob/f387cd2afec9419c8ee37694406ca490c3f34ee5/flake.nix#L2C17-L2C58

And here is some text describing the project that could be in the readme:

flake-compat/default.nix

Lines 1 to 6 in f387cd2

	# Compatibility function to allow flakes to be used by
	# non-flake-enabled Nix versions. Given a source tree containing a
	# 'flake.nix' and 'flake.lock' file, it fetches the flake inputs and
	# calls the flake's 'outputs' function. It then returns an attrset
	# containing 'defaultNix' (to be used in 'default.nix'), 'shellNix'
	# (to be used in 'shell.nix').

So it allows to create default.nix and shell.nix that run some parts of flake.nix. Likely should not be used by anyone, better extract common code into nix/ folder and call from both the flake and default.nix/shell.nix or put the code e.g. into shell.nix directly and call into it from flake.nix.

[eclairevoyant]

Yes, I agree that it's counterproductive when you can write more modular code. I did the same in https://github.com/feel-co/hjem/blob/main/default.nix.