Open stuartpb opened 7 years ago
The thing is, zip/tar/gz/targz files don't necessarily need to be extracted - but then we're in the world where we've got weird duck-typing logic and specs can have different behaviors that can't be predicted until the request completes.
Though I guess there's the "blobs require an as
" rule right now, and anything without that is interpreted as an archive, and that's probably OK? (archives don't have as
, they have slice.to
iirc)
It'd be nice if type could be implied by file extension in the source URL / filename (source or destination), and/or Content-Type on fetch (for sources), and/or some kind of duck typing (for
patch
bits).What'd be most crucial would be establishing the precedence - I'm thinking it'd look like (highest to lowest):