MP3 files present an interesting case study for metadata and a great starting point for Durham. Not that MP3s were the first files to use metadata but they have a lot of interesting conditions that need to be taken into consideration.

MP3 files are MPEG audio data. There's some metadata about the MPEG frames that can be derived from the file but cannot be modified. From Durham's perspective these properties can be read-only metadata.

Then we have id3v2 and id3v2 -- two metadata formats for MP3s that can exist on the same file. It's important to keep the information in both formats in synch as much as possible but the id3v1 format has severe limitations (the biggest properties are 30 characters maximum, for example) that can prevent this. The idea is to write the id3v1 tag to files for older/simpler applications that can't read the id3v2 tag.

Here's where Durham comes in: three different metadata formats being read but only two being written. Plus, the two being written should be written to the file at the same time for efficiency reasons instead of opening the file twice. Read-only properties (of course) are skipped and not written at all. Durham can manage all of this with a generalized metadata property API.

The plugins for the metadata formats MPEG, id3v1 and id3v2 will use the Durham API/interfaces to let Durham read and write for those formats. Durham will handle all of the file I/O -- the metadata format plugins just need to tell it where to read and write.

Third parties can add support for other file metadata formats by writing their own plugins.