Greg Pfeil

~The formatter failures are annoying – despite them being in Nix, I can’t replicate them locally on the same architectures, and the same checks are passing on `aarch64-*`.~ Nevermind –...

> It is not entirely clear to me that this fixes the problem we had. We never set `-funbox-strict-fields`; it's the `linear` package that sets it. > > Could you...

> Also, any idea why some tests require ignoring interface pragmas? I thought interface pragmas should never be ignored in practice (otherwise the plugin wouldn't be able to inline anything)....

Would it be possible to somehow apply `Client.Rep` and check that type? `Barbies.Dict` (and any other `Dict`-like that appears) should have `Constraint.Dict` as its `Rep`. Also, point of clarity –...

Should we mention this in the README, and perhaps set `-fno-unbox-strict-fields` in the `DynFlagsPlugin` (assuming I ever get around to adding it).

> We could potentially get the best of both worlds if we converted to Hask (as used for the plugin tests) in the first phase, then substitute Hask → cat...

I see this issue on macos as well. UCM version: ucm version: release/0.5.19 OS: macos 14.4 Shell: Zsh Terminal: Apple Terminal Unison installation: unison-nix overlay

Ah, I see that this is a consequence of `Pretty VersionRange`, and that set notation is just syntactic sugar in the parser, not maintained in the AST.

To help make modules more portable, is there some attribute that would return `/run/current-system/sw` in a nixosModule and `/run/system-manager` on a systemModule? I’d be happy to fix the hardcoding of...

Most of the commits here are part of #5154, which simplifies the flake on trunk. This PR consists of just a single commit. Also, there is one test that fails...