> There is another question waiting in the wings: How would we handle > this if you don't want to store the document byte stream as a blob in > the database, but leave it on the file system? > In the SQL-OS and I could just write a valuemapper that knows to > fetch the actual byte stream via the filename (which would be stored in > the database).. this uses the existing object store API. Is there another > solution?
Maybe we can leave this up to users? It is possible to store the filePath in database and have another non-persistent attribute for the blob. So use can handle loading the file in a getter. I would suspect that in this case it might be better to write some helper to do this instead of putting it to implementations of data stores. Giedrius.
