The ASD file is IN-PACKAGE ASDF, which is really not A Good Thing.
You could at least potentially damage a user's working ASDF by smashing some symbol in that package.
The comment at the header expresses a dislike for adding a new package. That's fine. If you don't put a package there, ASDF will load it into ASDF-USER.
So we should either:
1. replace ASDF with ASDF-USER in the IN-PACKAGE -- that's what I would favor, since it allows a programmer to interact with this definition in the editor or
2. simply remove the IN-PACKAGE form altogether, and ASDF will automagically put it into an appropriate package.
Best, r
- simply remove the IN-PACKAGE form altogether, and ASDF will
automagically put it into an appropriate package.
i've done this with some other cleanups and pushed the changes into the darcs repo.
let me know if there's anything else to be done.
On 7/11/16 Jul 11 -5:20 AM, Attila Lendvai wrote:
- simply remove the IN-PACKAGE form altogether, and ASDF will
automagically put it into an appropriate package.
i've done this with some other cleanups and pushed the changes into the darcs repo.
let me know if there's anything else to be done.
FWIW, I think the definition for the test op can be simplified.
It's enough to put
:in-order-to (test-op (test-op "iterate/tests"))
into the definition of iterate (the test-op has a built-in dependence on load-op for the same system), and then change the perform to
(defmethod asdf:perform ((op asdf:test-op) (c (eql (find-system ':iterate/tests)))) (funcall (intern "DO-TESTS" (find-package #+sbcl "SB-RT" #-sbcl "REGRESSION-TEST"))))
That isn't actually a necessary change, but it's a lot tidier. The current framework has the oddity that
(asdf:test-op "iterate/tests")
won't do anything.
It might even be possible to do away with the DEFMETHOD and put a :METHOD form in the defsystem for iterate/tests, but even I'm a little unsure about the syntax and semantics of that....
Best, r
FWIW, I think the definition for the test op can be simplified.
thanks Robert, pushed more patches implementing your suggestion (and i've retired the pg.lisp integration into an examples directory).
-
Attila Lendvai -
Robert P. Goldman