My workflow when it comes to development on lisp libraries, is to change into the project the launch a SLIME shell. This works great but how do I load systems from that project?
Well I used to use a snippet like this.
(push (truename (make-pathname :directory (sb-posix:getcwd))) asdf:*central-registry*)
This is not what most people seem to be doing, instead they are using
a local projects directory to host any local projects for example in
roswell that directory is
.roswell/local-projects. But often I find
that I might want to try and extend a system, or to hack on it, so I
like to have local copies.
I had experimented with creating symlinks into the local project dir, but in the end it has always left me unsatusfied.
While looking into better ways to do this I found that the
asdf:*central-registry* is not recommended as the way to add paths
to the load order. Instead it's recomeneded that you add
configuration files to
that will state where to find systems.
It does have one limitation though, you can't exceute anything, the files are loaded and treated as data.
But through some experimentation I found that read macros are honored,
so doing something like this will always add the current working
directory of any lisp to the load path. Just insert it into
;;; -*- Mode: Lisp; Syntax: COMMON-LISP; Base: 10 -*- (:directory #.(uiop/os:getcwd))