Hi, See http://uppertank.net/blog/archives/40 for a case where you'd like to have your mime type information stored and managed in an external repository. The way to do that with our current API would be to implement an alternative reader that reads all mime type information from the external repository and instantiates a MimeTypes registry with all that information.
The downside of this approach is that the type information is static and needs to be kept fully in memory. I'm not too concerned about the memory overhead (the registry can't be *that* big), but it would be nice to be able to update the mime type information without having to do full reloads of the registry. So, how about adding MimeTypeRegistry interface and making the current MimeTypes class be the default implementation of that interface? Then external repositories could implement the MimeTypeRegistry interface using on-demand loading of type information. Such a split would also make for a cleaner interface/implementation separation within Tika itself. BR, Jukka Zitting
