For the CI nerds in the room... I found

[marsh crater]

using the linter config means we're potentially busting cache too often, which i guess is an ok pain to live with for now

[split turret]

Didn't think of that. How would it bust cache?

(cleaning up then will share a link)

my CI PR is getting huge again 🙁

[marsh crater]

because the files aren't actually excluded from the dag, just from the linting process itself

so if lint excluded files change, the linters cache is still busted. Unless we have our linter module interpret the config and exclude them itself

[split turret]

Mmmm. True but, those files were already not filtered out before

We could always filter them out if we prefer

(assuming there's a way for the caller to apply ignore filters on the callee, which we need anyway as discussed)

[marsh crater]

Yeah exactly. I think its probably fine. Especially since we can handle it in the toolchain

[split turret]

Speaking of which, do you also want to build that @marsh crater, or something we can parallelize? I know you've got Kubecon stuff this week

[marsh crater]

which part was that? i'll join lounge in a bit, just getting to a stopping point with the module i'm working on

[split turret]

@marsh crater I'm going to use this thread to dump follow-up discoveries & thoughts. No response required

• Top-level source. Maybe it should be a *dagger.GitRepository instead of Directory, as a convention for toolchains. If we could handle that gracefully, it could make some stuff easier, like default ignore (could be natively handled by the Git API)