remove iogx

[tgunnoe]

ETCM-8233

iogx is unnecessarily complex for our use cases. nosys essentially only handles systems for you, rewriting the outputs. As such, it also fixes interfaces like 'nix flake show' where Input From Derivation (haskell.nix) would usually break the interface.

Other changes:

• Shells made into a single shell
• Updated nixci to omnix
• Cleanup outputs

Prereview checklist

• The CHANGELOG.md has been updated under the # Unreleased header using the appropriate sub-headings with links to the appropriate issues/PRs.
• All tests pass in CI
• PR was self-reviewed

[nrdxp]

might be nice to have this in a seperate script, perhaps somewhere in nix/scripts?

[tgunnoe]

Good idea. This is just a migration from how it was before (and legacy stuff from mlabs). And there's probably a non-nix way within the makefiles somewhere as well, so I'll just create a ticket to keep track of it.
In general, let scripts be useable outside of nix.

[nrdxp]

yeah I saw it wasn't any better before the change, just nice to split up longer than a few lines to its own file

[kpinter-iohk]

Nix shell doesn't seem to provide cabal. Not sure if I'm missing something.

$ nix develop
$ cabal
bash: cabal: command not found

[tgunnoe]

Nix shell doesn't seem to provide cabal. Not sure if I'm missing something.

$ nix develop
$ cabal
bash: cabal: command not found

Thanks, looks like iogx provided that while it made its shells. I'll add the toolchains missing from there

[kpinter-iohk]

After pulling:

1. cd offchain/ && make update-scripts fails to generate RawScripts.purs. This is because for some reason IO.withFile doesn't handle Unicode anymore, and there are › and ∷ characters in the output it's supposed to write. Seems like there was some magic in the previous version that fixed this. I'm okay with replacing these characters with their non-unicode equivalents too (> and ::).
2. After fixing the above issue with the use of non-Unicode chars, make build fails with this again: remove ctl builds and build everything using local tooling #98 (comment)

Minor comment:
I also noticed that on master we have cabal 3.12.1.0 and ghc 9.6.6, while on this we have cabal 3.10.3.0 and ghc 9.6.5.

[tgunnoe]

Thanks Krisztian. yeah, the tooling isn't correct yet. there was a whole chain of tooling that iogx would provide locally. I'm still working on it and will re-request your review once I get it right.

[tgunnoe]

@kpinter-iohk can you to see if all of your shell/build tooling is functional now?

[kpinter-iohk]

I see that the CI broke because script sizes changed. I wanted to look into this locally, however cabal build now fails when trying to resolve dependencies:

$ cabal build --minimize-conflict-set
Warning: Requested index-state 2024-07-16T05:49:34Z is newer than
'hackage.haskell.org'! Falling back to older state (2024-07-16T05:41:59Z).
Resolving dependencies...
Error: cabal: Could not resolve dependencies:
[__0] trying: trustless-sidechain-6.1.0 (user goal)
[__1] next goal: plutus-tx (dependency of trustless-sidechain)
[__1] rejecting: plutus-tx-1.30.0.0, plutus-tx-1.29.0.0, plutus-tx-1.28.0.0
(constraint from project config
/home/.../partner-chains-smart-contracts/onchain/cabal.project requires
^>=1.27)
[__1] trying: plutus-tx-1.27.0.0
[__2] next goal: plutus-core (dependency of trustless-sidechain)
[__2] rejecting: plutus-core-1.30.0.0 (conflict: plutus-tx =>
plutus-core^>=1.27)
[__2] skipping: plutus-core-1.29.0.0, plutus-core-1.28.0.0 (has the same
characteristics that caused the previous version to fail: excluded by
constraint '^>=1.27' from 'plutus-tx')
[__2] trying: plutus-core-1.27.0.0
[__3] next goal: cardano-crypto-class (dependency of trustless-sidechain)
[__3] rejecting: cardano-crypto-class-2.1.5.0, cardano-crypto-class-2.1.4.0,
cardano-crypto-class-2.1.3.0, cardano-crypto-class-2.1.2.0 (conflict:
pkg-config package libblst-any, not found in the pkg-config database)
[__3] rejecting: cardano-crypto-class-2.1.1.0 (conflict: plutus-core =>
cardano-crypto-class^>=2.1.2)
[__3] skipping: cardano-crypto-class-2.1.0.2, cardano-crypto-class-2.1.0.1,
cardano-crypto-class-2.1.0.0, cardano-crypto-class-2.0.0.1,
cardano-crypto-class-2.0.0.0.1, cardano-crypto-class-2.0.0 (has the same
characteristics that caused the previous version to fail: excluded by
constraint '^>=2.1.2' from 'plutus-core')
[__3] fail (backjumping, conflict set: cardano-crypto-class, plutus-core,
trustless-sidechain)
After searching the rest of the dependency tree exhaustively, these were the
goals I've had most trouble fulfilling: plutus-tx, plutus-core,
cardano-crypto-class, trustless-sidechain

[kpinter-iohk]

🎉 🥇