I have some classes that represent, for the most part, data that's deserialized from XML. They also have some behavior in them, because I don't want to suffer from an anemic domain model. These domain objects don't have any XML specific code in them, but they do have properties that correspond exactly to the tags and attributes in the XML.
Many of these classes' behaviors rely on resources loaded from disk, like images. For that, some properties / xml attributes are relative file paths to these resources. Now my problems begin. The absolute path to be constructed for these resources is based on a separate dependency - the root of the "project". Here's what I'm struggling with:
Should my domain objects store the full absolute path to their resources? Or just the relative paths? If they held the absolute path, then the classes that load them - my XML readers - will have to take a dependency on the project root in order to build the full path for them. If not, I'll have to make yet another layer of classes to handle that. I already have 3 layers, I don't know if I can handle any more!
Should my domain objects store their own paths for where they are on disk? It makes sense for them to know where their dependencies are in relation to them, but to know their own location just feels wrong to me intuitively. This is sort of unrelated to the other part of the question, but I think if I go the wrong way on this it could bite me later.