I'm not sure how to feel about this.

On one hand, it's an appealing way to manage dealing with different OS/Arch for a binary. On the other hand, this is not something I'd want to use in production, as it's very implicit. I'd rather explicitly call specific versions of installed binaries over implicitly call files installed in $DOTSLASH_CACHE (particularly if the binary is fetched on-demand).

This looks like it could be a useful tool in certain situations, such as shared/consistent dev environments. But generating the dotslash file (with hashes, urls, etc...) seems quite manual, which will limit adoption. In this way, it's not too different from traditional environmental module files, but it does automate a good chunk. If the main use-case is development, I'm not sure if this is better than devcontainers.

It seems like a useful tool, but I'm not sure where I'd actually use it. Maybe I'm missing the main use cases?

It's a tool for fetching things and making sure what you fetch is what it's supposed to be.

For facebook in particular I'd guess it's basically just a bootstrapping step for buck2 and other basic tools like that.

Containers are obviously tempting but I think they're basically a way of shoving complexity under the bed whereas the approach that dotslash nudges you towards is slightly hormetic in that it helps you manage the complexity while also acting a slight tax on it's growth.