Import inside flake.nix inputs?

Malix · May 28, 2026, 8:15am

starting to get off-topic, but the inputs themselves are not which causes confusion

the point about inputs not being fully turing complete too and thus making users rely on even more unsafe workarounds is a solid argument too :

github.com/NixOS/rfcs

Comment by Malix-Labs - [RFC 0193] TOML Flakes

master ← roberth:declarative-flakes

https://github.com/NixOS/rfcs/pull/193#issuecomment-3796693964 from @KiaraGrouws…tra: > if `inputs`' restrictions were lifted If all those restrictions were lifted, this RFC would be a non starter And those restrictions are stemming from preventing underlying problems such as those raised in https://github.com/NixOS/nix/issues/4945#issuecomment-868572873 : > to prevent arbitrarily complex, possibly non-terminating computations while querying flake metadata Also using IFD to compute flake inputs, which would break the whole philosophy of nix, and imo (and this is the whole debate around NixOS/nix#4945) sensical to prevent (at least by default) However, the point raised in https://github.com/NixOS/nix/pull/15132#issuecomment-4207131122 by @roberth : > Arguably it's more controlled than the alternative, which is that users have to cobble together input modifications using scripts that perform text replacements in flake.nix, or things along those lines. Is a strong argument; People really wanting to compute their flakes inputs are just doing it regardless, in an unsafer way. And other non-flakes solutions (I'm thinking of https://github.com/andir/npins) typically can allows for such behavior. In my opinion, the biggest pain point that RFC is solving is `flake.nix` not being nix but a subset of nix with a specific schema and thus misleading (what lead to the creation of NixOS/nix#4945) (also see https://github.com/NixOS/nix/issues/4945#issuecomment-4206680838) If "flakes inputs really shouldn't allow for non-trivial expressions": they should not have been primarily nix to begin with, and this RFC should have been the standard default; Else: NixOS/nix#15132 makes sense and this RFC shouldn't be accepted.