Consider URL's as file locations

[jlgelpi]

File path should be expanded to include remote locations, typically through a URL. The Apps will eventually download the raw data or activate a remote application (REST API , etc) to download any processed data. This goes along with the SAB comment in being prepared for a distributed storage.

[markmcdowall]

The file_path is just a str for where the file is located. From a phone call on Monday with Marco and Brian we were planning on having the VRE manage the uploading of the file. The VRE would then log the files location in the API. We the files need to be distributed this location can then be converted the the relevant path for FTP access.

Theoretically though this is just a path so it should be fine to define a URL instead of a directory.

Will keep this open as this is something that we can think about.

[jlgelpi]

I would prefer to think that any data is remote, even if it is on the VRE disk (Actually local data will be also available through a REST API). What I am using in the initial test on metadata are identifiers like repository:Identifier with a resolver from 'repository' to the actual REST url. (CURIE format if i'm not wrong). This notation would be valid anywhere.