Introduced by https://github.com/srid/haskell-flake/pull/253
Tangentially related: https://github.com/srid/haskell-flake/issues/192
But now that we have otherOverlays
I wonder if I should just get rid of custom
.
I can confirm that the order in which settings are applied is the issue. If I rename custom
to A_custom
, it works. This suggests that custom
doesn't work because it comes after buildFromSdist
in alphabetical order.
Srid said:
Tangentially related: https://github.com/srid/haskell-flake/issues/192
But now that we have
otherOverlays
I wonder if I should just get rid ofcustom
.
Hmm, we can't do that, because overlays don't compose when import
'ing.
Here's an idea: provide two settings:
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.https://github.com/srid/haskell-flake/issues/283
Need :eyes: and ideas from people.
Or,
settings.<name>.custom.initial
settings.<name>.custom.final
Or,
packages.<name>.drv
(computed from source
automatically, by default)settings.<name>.custom
(applies at the end, after other settings)Incidentally, I began thinking about why we even need separate packages
and settings
option groups. Like, why not this:
{
packages = {
hello_rust.drv = self'.packages.hello_rust
myhaskell.source = ./haskell;
myhaskell.setting.libraryProfiling = true;
aeson.source = "1.5.0";
};
}
This is nice, keeps all the info about a package in one-place.
Will be a big ass refactoring, will try from Australia next week.
Since this bug affects only non-existent packages, it is not high prio to fix I guess.
I could add the <package-name>.drv feature for non-existent packages, which shouldn’t be hard to do and not break anything
Will help publish this tutorial soon: https://github.com/nixos-asia/website/pull/24
Shivaraj B H said:
I could add the <package-name>.drv feature for non-existent packages, which shouldn’t be hard to do and not break anything
Feel free to give it a try.
Turns out we can fix this particular bug without implementing this proposal (which should be good on its own merits, though):
https://github.com/srid/haskell-flake/pull/284
srid on appreciate haskell-rust-ffi-template on master [⇣]
❯ g log -1
commit 1038ca8a87aae4e448bbd7ea6e1c41ee4e7acb2d (HEAD -> master)
Author: Shivaraj B H <[email protected]>
Date: Wed Mar 27 15:04:33 2024 +0530
ci: upgrade nixci
srid on appreciate haskell-rust-ffi-template on master [⇣]
❯ nix build --override-input haskell-flake github:srid/haskell-flake/settings-null --show-trace
warning: not writing modified lock file of flake 'git+file:///Users/srid/code/haskell-rust-ffi-template':
• Updated input 'haskell-flake':
'github:srid/haskell-flake/f9d17c3aa68e65529f424816c8b9346ae602d1de' (2024-02-06)
→ 'github:srid/haskell-flake/d6a9187919cbf087b99d88d154086a20fc1b72f1' (2024-03-28)
srid on appreciate haskell-rust-ffi-template on master [⇣] took 5s
❯
Shivaraj B H has marked this topic as unresolved.
seems like it is broken again: https://github.com/srid/haskell-flake/issues/283#issuecomment-2031339304
Does this fix it?
https://github.com/srid/haskell-flake/pull/298
Yes, it does.
here’s what I tried:
nix build github:shivaraj-bh/haskell-rust-ffi-template --override-input haskell-flake github:srid/haskell-flake/e20cdd30ad9201bedc4977fa55f7a4765d6a56ac
Last updated: Nov 21 2024 at 09:45 UTC