I want to understand precisely what Dagger considers when determining a cache hit/miss.
I've got a function that should be cached when there's no changes in a directory, and not cached otherwise. That directory is attached to a container, and that container (plus another container) are passed as arguments to that function. Those containers are created in a previous method with other inputs, which is why I think this isn't caching as I'd expect.
E.g.: with-config --dir . --workload-ver tag1 function-to-cache -> when --workload-ver changes (which is attached to a container with WithEnvVariable) the function to cache doesn't cache, it runs. Is there a way to write that function such that it doesn't matter which tag is pulled, only that if there are changes in the --dir . directory run, otherwise cache?
Trace: https://dagger.cloud/mjb/traces/e892c1d167c5541ef90273b87b54c9c6. Of the two long-running functions that ran without caching, the first should only care about the directory contents, the second should run when the --workload-ver changes as usual.
