Commonly, one wants to extend ASDF with new operation and component classes.
But the support in ASDF for referencing such classes all involves automagically interpreting keyword symbols (and some unqualified symbols?) in the ASDF package.
If I understand this correctly, this leaves the ASDF extender with the alternatives of either jamming their new symbols into the ASDF package late, and exporting them, or having them live in a new package, which is cleaner, but relegates these new components to second-class status.
Consider the following
(defpackage my-system-def (:use common-lisp asdf))
(in-package :my-system-def)
(defsystem my-system :defsystem-depends-on ("my-asdf-extension") :components ((:my-new-file "file1")))
I don't seem to be able to define MY-NEW-FILE in my-asdf-extension and export it into my-system-def, because my-system-def is already defined by the time my-asdf-extension is loaded. That means I pretty much have to either muck around in the ASDF package (which doesn't scale to multiple different people extending ASDF) or use side effects:
(asdf:load-system "my-asdf-extension")
(defpackage my-system-def (:use common-lisp asdf my-asdf-extension))
(in-package :my-system-def)
(defsystem my-system :components ((:my-new-file "file1")))
Am I right about this? If I am, is there some way to better handle this? Or do we need to rethink how :defsystem-depends-on works?