How Tagging Works

• Tags shall be stored on each file as extended attributes named user.tagids
  • On windows this is stored as an alternate stream with the same name
  • Containing a string of unsigned 16 bit integers (not encoded)
  • First id is a namespace identifier (more on this later)
• A UTF8 file named .tagnames may exist at any directory as a way to translate the tag IDs and to allow instant renaming
  • This is not a directory extended attribute because not every system can safely store large attributes
  • The first two lines of the file shall serve as its header
    • The first shall serve as the file magic number
    • The second shall be a random hex number, serving as a namespace identifier to detect out of place files
  • The body of the file shall be a list of tag names
    • Delimited by new line characters
  • The file does not necessarily need to be in the opened directory
    • program shall bubble up the directory tree until it finds the file
    • the search ends when we arrive at the home of the user or a mount point
• The application shall detect "out of place" files i.e. files whose namesapce IDs are foreign to the nearest parent
  • We don't store the path to the namespace because it can leak meta-data about the directory
  • Disregarding the security issue, storing the path doesn't mean the namespace file won't be moved, so it's not functionally better
• Both tag and namespace IDs starts from 1 to prevent falsy values & serialization errors from silently creeping in