meta(library usage): tracking issue for using Foundry as a library

[nhtyy]

Component

Forge, Cast, Anvil, Chisel

Describe the feature you would like

As Foundry has become the flagship testing engine for the Solidity/EVM community, more and more people will rely on the functionality it offers on the server.

Currently however there are some pain points, some examples:

• Reusing artifacts in Forge: If you store the artifacts in some external database, the MultiContractRunnerBuilder could be more generic I think

  • Linking can also silently fail, if you don't have correct artifact ids (err on any non linked test files?)

• Injecting env vars in forge: Currently if you want to run tests on the server you must fork and disable/reroute vm.env.

  • also cant disable cheats
  • i propose feature flagging reading env from the cheat codes somehow

• Coverage: You need to just run the bin because so much happens in binary, which would need to be duplicated if you want lcov

Additional context

No response

[nhtyy]

@zerosnacks happy to make these changes and expand on this if you think this is an effort worth tracking

I think If we can just maintain the high level abstractions across anvil/foundry/chisel then most people would not need to fork to safely run on server

[zerosnacks]

Hi @nhtyy thanks for your suggestion, just so I have a bit more context - could you expand on possible use cases you see for using Foundry on a (public) server that runs untrusted code?

[nhtyy]

@zerosnacks forsure! One example is something Ive worked on personally myself which is audit wizard, which allows you to run forge tests against any repo.

I know theres another one that came out recently, that also allows running tests but i cant find it now

[nhtyy]

Ill update this issue soon with new subissues for these tasks, and see if i can find anything else

[grandizzy]

I know theres another one that came out recently, that also allows running tests but i cant find it now

there's https://getrecon.xyz/ too for invariants