I think it would be much easier for Neuro to specify file operations in bulk if she needs to do multiple file operations at once.
My first thought was temporarily emulating a filesystem in memory using memfs - turns out, in order for it to be environment-agnostic, we'd need to build an adapter class for it and perform value injections for tree-shaking purposes.
(also imports from that package are very slightly broken which means we need to patch it using pnpm patch, which I will warn, can be pretty hellish, but it depends on the package and how it's structured after build.)
My second thought was to make our own step-by-step runner for this purpose, but that would probably be a lot of effort.
Open to implementation suggestions.
I think it would be much easier for Neuro to specify file operations in bulk if she needs to do multiple file operations at once.
My first thought was temporarily emulating a filesystem in memory using
memfs- turns out, in order for it to be environment-agnostic, we'd need to build an adapter class for it and perform value injections for tree-shaking purposes.(also imports from that package are very slightly broken which means we need to patch it using
pnpm patch, which I will warn, can be pretty hellish, but it depends on the package and how it's structured after build.)My second thought was to make our own step-by-step runner for this purpose, but that would probably be a lot of effort.
Open to implementation suggestions.