On Tue, Nov 3, 2015 at 9:56 PM, Eli Zaretskii wrote: > > > From: Dmitry Gutov > > > I don't know which solution to pick, and I believe I don't know the > > problem well enough. We probably don't want to venture into Java-land > > here, and encapsulate any type of field into its separate data type, > > with its comparison function. But, well, we could do that. > > I hoped Juanma will pick up the gauntlet. At this point, I don't even know what is being proposed. A file-name data type (in the cl-defstruct sense) with appropriate canonicalization and comparison functions?