-
Notifications
You must be signed in to change notification settings - Fork 20
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
custom
option is broken for *new* packages
#283
Comments
…ad of custom custom is broken, see: srid/haskell-flake#283 Perhaps, custom is not the right way to add non-haskell pkg deps, it makes more sense to add them under otherOverlays
It probably doesn’t make sense to add non-haskell pkg dependencies in custom, it should only be used for managing haskell packages, which it is meant for. It made more sense to move |
If we stick to that, we needn’t worry about the order in which the setting is applied, do we? |
People use Regarding,
(The reason for preventing merging of overlay lists here is due to nondeterministic semantics of list merge) |
custom
option is brokencustom
option is broken for *new* packages
It is breaking again after #286 Reproducer: nix build github:shivaraj-bh/haskell-rust-ffi-template --override-input haskell-flake github:srid/haskell-flake/51bcbc27194e734a1524ac3d36e1539fba9a3baa |
@shivaraj-bh It is breaking, but it is not the same error. From the issue description:
What I'm seeing when using
Could you open a separate bug report? |
Actually, wait. Let me figure out what the issue is first. |
Triggered (if not introduced) by #253 which made it so that
buildFromSdist
is applied (alphabetically) before thecustom
option, and for some reasonx: x
in Nix is not lazily propagating the argument.Repro & Details
https://nixos.zulipchat.com/#narrow/stream/413949-haskell-flake/topic/.60custom.60.20setting.20stopped.20working
Proposed design change
Replace
custom
with two setting options:settings.<name>.init : drv
: assigns the initial value before settings are appliedsettings.<name>.final : drv -> drv
: transforms the final drv after the other settings are applied.Open question: putting these two options in the same namespace as other settings is a bit of hack, because they are semantically different. Main questions: a) Is this the best we can do as proposal? b) if so, where should these options live?
(See Zulip for further proposals/ refinements/ ideas)
The text was updated successfully, but these errors were encountered: