Including ansi-test

[ian.t...@gmail.com]

I wonder about inclusion of ansi-test:

https://gitlab.common-lisp.net/ansi-test/ansi-test

First, I wonder if anyone knows of similar efforts.  Second, I wonder if folks think it makes sense to include it in cl-test-grid, or if some other implementation conformance project should take it on.

[Anton Vodonosov]

29.04.2016, 04:51, "ian.t...@gmail.com" <ian.t...@gmail.com>:

> I wonder about inclusion of ansi-test:
>
> https://gitlab.common-lisp.net/ansi-test/ansi-test
>
> First, I wonder if anyone knows of similar efforts.  Second, I wonder if folks think it makes sense to include it in cl-test-grid, or if some other implementation conformance project should take it on.

I don't know of similar efforts.

I would like ansi-test to be included, and thought about it. It would require creating ASDF system for ansi tests and including them into Quicklisp.
So that we could do:

(ql:quickload :ansi-test)
;; And them call some function which runs tests and return result.
;; ansi-test seems to use the RT framework, so cl-test-grid will do
(run-rt-test-suite)

[Mark Evenson]

On Friday, April 29, 2016 at 3:58:37 AM UTC+2, Anton Vodonosov wrote:

29.04.2016, 04:51, "ian.t...@gmail.com" <ian.t...@gmail.com>:
> I wonder about inclusion of ansi-test:

I would like ansi-test to be included, and thought about it. It would require creating ASDF system for ansi tests and including them into Quicklisp.
So that we could do:

  (ql:quickload :ansi-test)
  ;; And them call some function which runs tests and return result.
  ;; ansi-test seems to use the RT framework, so cl-test-grid will do
  (run-rt-test-suite)

ANSI-TEST uses a fair number of CL:LOAD "tricks" for various parts of its tests that make encapsulation in ASDF a reasonably non-trivial problem, although the refactoring done by Daniel Kochmański when he moved the project into the common-lisp.net gitlab instance over the past year alleviates that somewhat.

As [I pointed out to the ANSI-TEST mailing list][1], ABCL contains an [ASDF definition][2] of the ANSI-TEST suite to the point that ASDF:TEST-SYSTEM will work, although it is notably missing an enumeration of the actual file components that it invokes.  The ASDF definition was developed against ASDF2, so could stand a suitable skeptical refactoring with respect to ASDF3 abstractions.  Were there to be work towards developing an actual ASDF encapsulation with ANSI-TEST, I would be interesting in contributing the migration and testing of ABCL to the new abstraction.

[1]: https://mailman.common-lisp.net/pipermail/armedbear-devel/2015-September/003607.html
[2]: http://abcl.org/trac/browser/trunk/abcl/abcl.asd#L76