Thanks for the updates. AFAIK Xach is uninterested in bundling a newer version of ASDF. I regret this, but there's nothing I can do about it. It would be great to get a more up-to-date version of ASDF bundled with CLISP; if there's anything that keeps that from happening, LMK, and I will try to help and coordinate with the CLISP team.

As I said, anything that the CLISP team could do to support ASDF testing would be welcome. I am somewhat knowledgeable about the CL Foundation Docker images, and have managed to roll new ones for SBCL, Allegro, etc. for testing purposes, so I could help with Dockerizing CLISP if there's anyone who's interested.

On 21 Nov 2024, at 12:15, Alexandru Popa wrote:

Hi Robert and sorry for not testing latest ASDF. The problem actually was detected during playing with QuickLisp, and I was under impression it has a relatively recent version of ASDF.

Tested with ASDF-master (https://gitlab.common-lisp.net/asdf/asdf), version 3.3.7.2, :HAIKU, :OS-HAIKU, :UNIX and :OS-UNIX are all present, which is probably good.

Tested with ASDF from QuickLisp (https://www.quicklisp.org/beta/), version 3.2.1, :UNIX and :OS-UNIX are present, :HAIKU is removed.

Tested with ASDF from CLIPS (https://gitlab.com/gnu-clisp/clisp), version 3.2.0, :UNIX and :OS-UNIX are present, :HAIKU is removed.

So, the solution is just using a more recent ASDF version. I will communicate this to both CLISP and QuickLisp teams. I will also ask Haiku and CLISP teams to provide a docker image or testing.

Thank you,

Alexandru

чт, 21 нояб. 2024 г. в 17:45, Robert Goldman <rpgoldman@sift.info>:

Please check [this merge request])(https://gitlab.common-lisp.net/asdf/asdf/-/merge_requests/147) and the corresponding ASDF issue.

Is CLISP bundling an up-to-date version of ASDF?

Please let me know, and if the above do not address the issue, open a new issue on the ASDF gitlab site and I will see about getting this fixed and issuing a bug fix release.

It would be very helpful if the clisp community could point me at a Docker image for Haiku + clisp so that I could put this configuration into test.

Indeed, it would be very helpful to have a CLISP docker image for linux kept updated for test purposes. The lack of an official release version hampers my testing substantially; so much so that I have dropped clisp from the test suite. It's impossible to know what to test when there hasn't been an official release in 14 years.

On 21 Nov 2024, at 5:26, Alexandru Popa wrote:

Haiku is an operation system different from Windows, Linux or *BSD. Still,
in many respects it can be considered Unix-like, which is especially true
when porting software to it.

CLISP was recently ported to Haiku. In the *features*, CLISP on Haiku
defines both :UNIX (i.e. Unix-like) and :HAIKU, and this is probably the
right decision. However, after (require "asdf"), *features* are changed,
:HAIKU is removed and :OS-UNIX is added. This is certainly something which
needs to be fixed. The desired solution would be to keep both :UNIX, :HAIKU
and to add :OS-HAIKU.

Here is the analysis from CLISP development list:
```
(defun detect-os ()
"Detects the current operating system. Only needs be run at
compile-time, except on ABCL where it might change between FASL compilation
and runtime."
(loop* :with o
:for (feature . detect) :in '((:os-unix . os-unix-p) (:os-macosx
. os-macosx-p) …
(:haiku . os-haiku-p))
:when (and (or (not o) (eq feature :os-macosx)) (funcall detect))
:do (setf o feature) (pushnew feature *features*)
:else :do (setf *features* (remove feature *features*))
:finally
(return (or o (error "Congratulations for trying ASDF on an
operating system~%~ that is neither Unix, nor Windows, nor Genera, nor even
old MacOS.~%Now you port it.")))))

That is somewhat brittle code that side-effects *FEATURES*. It contains a
special bypass to allow :OS-MACOSX to live there beside :OS-UNIX, but
there’s nothing equivalent for Haiku. Whether Haiku is considered a UNIX or
not I won’t debate.

I call such code /brittle/ because there’s an undocumented (non-explicitly
mentioned) dependency on element order in some innocuous list: If
:os-macosx were before :os-unix in the A-list, that code would not set –
and rather delete – :OS-UNIX in *features*.

I would have appreciated a tiny comment like

:for (feature . detect) :in '(…
(:os-unix . os-unix-p) (:os-macosx . os-macosx-p) ; Beware,
unix must come first!
'''

Alexandru Popa

Robert P. Goldman
Research Fellow
Smart Information Flow Technologies (d/b/a SIFT, LLC)

319 N. First Ave., Suite 400
Minneapolis, MN 55401

Google Voice: (612) 326-3934
Cell: (612) 384-3454
Email: rpgoldman@SIFT.net

Robert P. Goldman
Research Fellow
Smart Information Flow Technologies (d/b/a SIFT, LLC)

319 N. First Ave., Suite 400
Minneapolis, MN 55401

Google Voice: (612) 326-3934
Cell: (612) 384-3454
Email: rpgoldman@SIFT.net