This project implements a single (heavily commented) file, lens.erl
which implements functional lenses in Erlang. I've been wanting to
check out the implementation of this for some time now, and Spawnfest
2012 was as good as any occasion.
A lens is a generalized accessor/mutator pattern in a functional
persistent setting. It allows us to chain and compose such accessors,
much like the normal foo.bar.baz() chaining in most OO languages. It
also starts a library for producing ways to "dig" into structures much
like xpath queries in XML. But on the contrary it also allows for
functional update and alteration of the same structures.
The intuition of a lens is as a lens in a camera. When we view a structure through the lens, we get an abstract view of what lies behind, twisted by the lens. This allows us to manipulate the structure behind the lens by casting it in a different light. We can pull the structure through the lens, operate on it, and then push it back through the lens to obtain what it originally was, but changed.
This project should be seen mostly as exploration of the concept. It is not yet ready for drop-in into large projects. But it can over time be made into such.
Finally, you will need eqc-mini. All the code is run through Erlang
QuickCheck in order to make sure we have a somewhat firm ground on
which to stand. That is, all our lens hacking is probalistically
checked for correctness. We also use eunit for certain simpler tests
in the code where we want to verify the point query of a given result.
If you have access to EQC mini you should be able to run lens:t() in
order to check the module. It runs both the eunit and EQC suites.