Re: Custom channel is making guix pull fail

[Luis Felipe]

On 13/09/24 17:14, Luis Felipe wrote:
> Hi,
>
> Could you help me find out what I'm doing wrong here, please? :)
>
> I have a custom channel 
> (https://codeberg.org/luis-felipe/guix-channel-x) that had been 
> working fine for a very long time, but now running "guix pull" fails 
> with the following error:
>
> #+begin_quote
> building /gnu/store/3aghnb3nbxmgl5f2fsjchz27iwsdz0rl-luflac-x.drv...
> -builder for 
> `/gnu/store/3aghnb3nbxmgl5f2fsjchz27iwsdz0rl-luflac-x.drv' failed to 
> produce output path
> `/gnu/store/zpmgrl9sz83ininzainjs9kjk25wymsw-luflac-x'
> build of /gnu/store/3aghnb3nbxmgl5f2fsjchz27iwsdz0rl-luflac-x.drv failed
> View build log at 
> '/var/log/guix/drvs/3a/ghnb3nbxmgl5f2fsjchz27iwsdz0rl-luflac-x.drv.gz'.
> cannot build derivation 
> `/gnu/store/99w71gp2rxzb7g1ij62qag769bryxwzi-profile.drv': 1 dependencies
> couldn't be built
> guix pull: error: build of 
> `/gnu/store/99w71gp2rxzb7g1ij62qag769bryxwzi-profile.drv' failed
> #+end_quote
>
> And the referenced build log reads as follows:
>
> #+begin_quote
> (repl-version 0 1 1)
> guix repl: warning: '%gnu-build-system-modules' is deprecated,
> use '%default-gnu-imported-modules' instead
> WARNING: (guix build guile-easy-build-system):
> imported module (guix build utils) overrides core binding `delete'
> (exception match-error (value "match") (value "no matching pattern") 
> (value ()))
> #+end_quote

Problem solved. And for the sake of closure:

Since I couldn't find the exact problem using guix c6ff1d6, I decided to 
exclude my channel and upgrade my system (now guix 0feeac3). This time, 
importing my build system module in a guix repl consistently reported a 
similar problem to the one displayed when performing "guix pull" in my 
initial message:

#+begin_src shell
❯ guix repl -L .
scheme@(guix-user)> (use-modules (guix build guile-easy-build-system))
While compiling expression:
Throw to key `match-error' with args `("match" "no matching pattern" ())'.
#+end_src

Great, because that way I could start debugging that module locally.

The problem was in the definition of the standard phases of my 
"guile-easy" build system:

#+begin_src scheme
(define %standard-phases
  (modify-phases gbs:%standard-phases
    (delete 'install-documentation)
    (delete 'compress-documentation)
    (delete 'delete-info-dir-file)
    (delete 'validate-documentation-location)

    (add-after 'set-paths 'add-output-to-guile-load-paths
      add-output-to-guile-load-paths)
    (add-after 'build 'check check)
    ;; These two below were responsible for the "match-error":
    (add-after 'install 'install-scripts install-scripts)
    (add-after 'install 'install-info-manual install-info-manual)))
#+end_src

The last two phases, "install-scripts" and "install-info-manual", were 
added after the "install" phase, but there is no such phase: The phases 
of my "guile-easy" build system inherit from "guile"'s build system 
phases (referred to as "gbs:%standard-phases" in the previous code), 
which in turn inherit from the "gnu"'s build system phases, but the 
guile build system explicitly deletes the "install" phase (among other 
phases).

Apparently, in this version of Guix referring to non-existing phases 
breaks the code. This wasn't the case in Guix c81ed09, for example, 
where I can build packages that use the guile-easy build system just fine.

So, the solution was to change the standard phases definition so that 
the faulty phases refer to phases that actually exist:

#+begin_src scheme
(define %standard-phases
  (modify-phases gbs:%standard-phases
    (delete 'install-documentation)
    (delete 'compress-documentation)
    (delete 'delete-info-dir-file)
    (delete 'validate-documentation-location)

    (add-after 'set-paths 'add-output-to-guile-load-paths
      add-output-to-guile-load-paths)
    (add-after 'build 'check check)
    (add-after 'check 'install-scripts install-scripts)
    (add-after 'check 'install-info-manual install-info-manual)))
#+end_src

On a side note, building a package that uses my faulty build system in 
my new Guix, a useful backtrace is printed, but it seems to me that 
there might be a defect in the Guix code there:

#+begin_src shell
❯ guix build -L . guile-documenta
guix build: warning: failed to load '(guix build guile-easy-build-system)':

Backtrace:
In guix/store.scm:
   1330:8 19 (call-with-build-handler #<procedure 7f7edc8811e0 at g?> ?)
In guix/scripts/build.scm:
    584:2 18 (_)
In srfi/srfi-1.scm:
   673:15 17 (append-map _ _ . _)
   586:17 16 (map1 ((argument . "guile-documenta") (build-mode . 0) ?))
In guix/scripts/build.scm:
   604:31 15 (_ _)
In gnu/packages.scm:
    485:2 14 (%find-package "guile-documenta" "guile-documenta" #f)
    365:6 13 (find-best-packages-by-name _ _)
   295:56 12 (_ "guile-documenta" _)
In unknown file:
          11 (force #<promise #<procedure 7f7edc8dc040 at gnu/packag?>)
In gnu/packages.scm:
   242:33 10 (fold-packages #<procedure 7f7edc6f35f0 at gnu/package?> ?)
In guix/discovery.scm:
   158:11  9 (all-modules ("." ("/gnu/store/03iqzmjsha60m3fk?" . #)) ?)
In srfi/srfi-1.scm:
   460:18  8 (fold #<procedure 7f7ed6611860 at guix/discovery.scm:1?> ?)
In guix/discovery.scm:
   145:31  7 (_ "." ())
    115:5  6 (scheme-modules _ _ #:warn _)
In srfi/srfi-1.scm:
   691:23  5 (filter-map #<procedure 7f7ed6611720 at guix/discove?> . #)
In guix/discovery.scm:
   123:24  4 (_ . _)
In guix/ui.scm:
    482:5  3 (warn-about-load-error "./guix/build/guile-easy-build-?" ?)
In unknown file:
           2 (display-error #f #<output: file /dev/pts/1> "match" "?" ?)
In ice-9/boot-9.scm:
  1685:16  1 (raise-exception _ #:continuable? _)
  1685:16  0 (raise-exception _ #:continuable? _)

ice-9/boot-9.scm:1685:16: In procedure raise-exception:
Wrong number of arguments to #<procedure display-error (_ _ _ _ _ _)>
#+end_src

It seems to me that "guix/ui.scm" tries to report more useful 
information about the problem loading "guile-easy-build-system" but then 
"display-error" complains about not getting the right number of 
arguments (?).

Anyways, my channel is working again, problem solved (or so it seems :)).

OpenPGP_0x0AB0D067012F08C3.asc
Description: OpenPGP public key

OpenPGP_signature.asc
Description: OpenPGP digital signature