We've now got two ways to populate an ocfl object - import a directory, and build a directory with a callback which gets passed the deposit target directory. Both of these require the whole object contents to be in place before running the hash algorithm over them to build the manifest.
Third alternative: add a file at a time to an object, hashing as we go. This will be useful for using ocfl objects as data capture targets and also for things like collaboration. Issues to work out - do we have a commit operation? How do we distinguish which version a new file is going into?
We've now got two ways to populate an ocfl object - import a directory, and build a directory with a callback which gets passed the deposit target directory. Both of these require the whole object contents to be in place before running the hash algorithm over them to build the manifest.
Third alternative: add a file at a time to an object, hashing as we go. This will be useful for using ocfl objects as data capture targets and also for things like collaboration. Issues to work out - do we have a commit operation? How do we distinguish which version a new file is going into?