See bootc-dev/bootc#1314 (comment)
In live chat Allison was thinking we could at least pick a default hash value; that would very clearly help in reducing typing. Maybe we could make it a repository build time option?
But I still have a medium-strong opinion (as came up in the original PR #105 ) that it'd be better to just do dynamic dispatch in general.
There's variants here; e.g. we might keep generics for e.g. pub struct FileSystem.
I dunno though, I meant we already heap allocate for each filename, extended attribute and inline content, are we really winning by inlining RegularFile::External::ObjectId?
See bootc-dev/bootc#1314 (comment)
In live chat Allison was thinking we could at least pick a default hash value; that would very clearly help in reducing typing. Maybe we could make it a repository build time option?
But I still have a medium-strong opinion (as came up in the original PR #105 ) that it'd be better to just do dynamic dispatch in general.
There's variants here; e.g. we might keep generics for e.g.
pub struct FileSystem.I dunno though, I meant we already heap allocate for each filename, extended attribute and inline content, are we really winning by inlining
RegularFile::External::ObjectId?