[bug#62307] [PATCH core-updates 00/15] Update Hurd and fix build failures

[Ludovic Courtès]

Josselin Poiret <dev@jpoiret.xyz> skribis:

> * gnu/packages/gcc.scm (gcc-11): Add ugly conditional patching, to avoid a
> world-rebuild.
> * gnu/packages/cross-base.scm (cross-gcc):
> * gnu/packages/commencement.scm (gcc-boot0): Pass the patch there as well.
> * gnu/packages/patches/gcc-11-libstdc++-hurd-libpthread.patch: New file.
> * gnu/local.mk (dist_patch_DATA): Register it.

[...]

> +      ;; FIXME This patch is not in the origin because it is conditionally
> +      ;; applied, to avoid rebuilds at the end of the core-updates cycle.
> +      ;; Move to gcc-11's origin ASAP.
> +      ((target-hurd? (or (%current-target-system)
> +                         (%current-system)))
> +       (list patch-boot0 (search-patch 
> "gcc-11-libstdc++-hurd-libpthread.patch")))
> +      (#t '())))))
>  
>  (define perl-boot0
>    (package
> diff --git a/gnu/packages/cross-base.scm b/gnu/packages/cross-base.scm
> index 2959616af6..30bf1c6bd0 100644
> --- a/gnu/packages/cross-base.scm
> +++ b/gnu/packages/cross-base.scm
> @@ -319,14 +319,21 @@ (define* (cross-gcc target
>                     ("libc" ,libc))
>                   `(,@inputs
>                     ("mingw-source" ,(package-source mingw-w64)))))
> -            (libc
> +            (else
>               `(,@inputs
> -               ("libc" ,libc)
> -               ("libc:static" ,libc "static")
> -               ("xkernel-headers"       ;the target headers
> -                ,@(assoc-ref (package-propagated-inputs libc)
> -                             "kernel-headers"))))
> -            (else inputs)))))
> +               ;; FIXME Kludge to only apply the patch for hurd to avoid 
> rebuilds
> +               ;; remove ASAP, along with the accompanying changes.
> +               ,@(if (target-hurd? target)
> +                     `(("patch" ,patch)
> +                       ("_" ,(search-patch 
> "gcc-11-libstdc++-hurd-libpthread.patch")))
> +                     '())
> +               ,@(if libc
> +                     `(("libc" ,libc)
> +                       ("libc:static" ,libc "static")
> +                       ("xkernel-headers" ;the target headers
> +                        ,@(assoc-ref (package-propagated-inputs libc)
> +                                     "kernel-headers")))


[...]

> +     ;; FIXME We conditionally apply a patch passed in native-inputs, to 
> avoid
> +     ;; rebuilds at the end of the core-updates cycle.  Move this to a 
> regular
> +     ;; patch ASAP.
> +     ((target-hurd? (or (%current-target-system)
> +                        (%current-system)))
> +      (substitute-keyword-arguments (package-arguments gcc-8)
> +        ((#:phases phases)
> +         `(modify-phases ,phases
> +            (add-after 'unpack 'patch-libpthread-hurd
> +              (lambda* (#:key inputs native-inputs #:allow-other-keys)
> +                (define patch (search-input-file (or native-inputs inputs) 
> "/bin/patch"))
> +                (define the-patch
> +                  (cdr
> +                   (car
> +                    (filter (lambda (input)
> +                              (string-suffix? 
> "gcc-11-libstdc++-hurd-libpthread.patch"
> +                                              (cdr input)))
> +                            (or native-inputs inputs)))))
> +                (invoke patch "--verbose"
> +                        "--force" "--no-backup-if-mismatch"
> +                        "-p1" "--input" the-patch)))))))
> +     (#t (package-arguments gcc-8))))
> +   (native-inputs
> +    (cond
> +     ((target-hurd? (or (%current-target-system)
> +                        (%current-system)))
> +      (modify-inputs (package-native-inputs gcc-8)
> +        (prepend (module-ref (resolve-interface '(gnu packages base)) 'patch)
> +                 (search-patch "gcc-11-libstdc++-hurd-libpthread.patch"))))
> +     (#t (package-native-inputs gcc-8))))

So, we should discuss with others whether we can instead do another
round of rebuilds.  I know, I know, but this is really quite complicated
and better avoided.

Side note: please use ‘if’ rather than ‘cond’ when there are only two
possibilities.

Ludo’.