What’s in a Nix store path

fzakaria · March 29, 2025, 12:31am

bjth · March 29, 2025, 1:14am

Cool, any guesses as to why it has to go through so many hashing and compressing? I’m curious to the historical reasoning behind it.

Ericson2314 · March 29, 2025, 1:40am

It could definitely be simpler. But the basic motivation for why the normalization is so changing fixed-output derivations don’t cause mass-rebuilds.

Demystifying Nix is pretty fun but there is definitely a lack of documentation beyond the thesis for how it all works.

We know! I started writing it up already:

github.com

NixOS/nix/blob/3f13cc0f8739a1a70e52326af9ec49cee536128a/doc/manual/source/store/derivation/outputs/input-address.md?plain=1

# Input-addressing derivation outputs

[input addressing]: #input-addressing

"Input addressing" means the address the store object by the *way it was made* rather than *what it is*.
That is to say, an input-addressed output's store path is a function not of the output itself, but of the derivation that produced it.
Even if two store paths have the same contents, if they are produced in different ways, and one is input-addressed, then they will have different store paths, and thus guaranteed to not be the same store object.

<!---

### Modulo fixed-output derivations

**TODO hash derivation modulo.**

So how do we compute the hash part of the output path of a derivation?
This is done by the function `hashDrv`, shown in Figure 5.10.
It distinguishes between two cases.
If the derivation is a fixed-output derivation, then it computes a hash over just the `outputHash` attributes.

If the derivation is not a fixed-output derivation, we replace each element in the derivation’s inputDrvs with the result of a call to `hashDrv` for that element.

This file has been truncated. show original

Anyone that wants to write docs is high encouraged to beat me to finishing that

fzakaria · March 29, 2025, 2:05am

@Ericson2314 probably the fingerprint pre-hashing, and compress is finally enough.

All that’s needed for mass rebuilds afaik, is replacing the path with either the CA path (for FOD) or the input path modulo FOD.

I think having gone through this the multiple hashes (hash of hash) and compress to 20 bytes seem unnecessary … I think

I just wrote this and all it did was open up more questions

llakala · March 29, 2025, 2:25am

I’d be interested to see a followup explaining how the content-addressed experimental feature plays into all this!