Re: [PATCH 2/3] coreutils: import prebuilt manpages from Gentoo

[Mark Hatle <mark.hatle@windriver.com>]

On 2/10/15 1:19 PM, Paul Gortmaker wrote:
> As can be seen here:
> 
> http://lists.gnu.org/archive/html/coreutils/2014-11/msg00001.html
> 
> other people have run into the headache of trying to cross compile
> coreutils, with the "help2man" problem (assumes host can run the
> target binaries with "--help" arg to collect data for manpage.)
> 
> And since "current wisdom is this is working as intended" we are
> largely left with no choice but to use the same solution and
> abandon trying to generate the man pages at build time.  So here
> we import prebuilt manpages.
> 
> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
> ---
>  meta/recipes-core/coreutils/coreutils_8.23.bb | 15 ++++++++++++---
>  1 file changed, 12 insertions(+), 3 deletions(-)
> 
> diff --git a/meta/recipes-core/coreutils/coreutils_8.23.bb b/meta/recipes-core/coreutils/coreutils_8.23.bb
> index be68a310311a..5454a43527cc 100644
> --- a/meta/recipes-core/coreutils/coreutils_8.23.bb
> +++ b/meta/recipes-core/coreutils/coreutils_8.23.bb
> @@ -12,14 +12,17 @@ DEPENDS_class-native = ""
>  
>  inherit autotools gettext texinfo
>  
> -SRC_URI = "${GNU_MIRROR}/coreutils/${BP}.tar.xz \
> +SRC_URI = "${GNU_MIRROR}/coreutils/${BP}.tar.xz;name=tarball \
> +           http://distfiles.gentoo.org/distfiles/${BP}-man.tar.xz;name=manpages \
>             file://man-decouple-manpages-from-build.patch \
>             file://remove-usr-local-lib-from-m4.patch \
>             file://fix-selinux-flask.patch \
>            "
>  
> -SRC_URI[md5sum] = "abed135279f87ad6762ce57ff6d89c41"
> -SRC_URI[sha256sum] = "ec43ca5bcfc62242accb46b7f121f6b684ee21ecd7d075059bf650ff9e37b82d"
> +SRC_URI[tarball.md5sum] = "abed135279f87ad6762ce57ff6d89c41"
> +SRC_URI[tarball.sha256sum] = "ec43ca5bcfc62242accb46b7f121f6b684ee21ecd7d075059bf650ff9e37b82d"
> +SRC_URI[manpages.md5sum] = "ce49bc8360e16afe67581d136606efc7"
> +SRC_URI[manpages.sha256sum] = "d1b16e74b3dba4f313aeebae2c69da205f7d2e6e1c2aa8d39026c860d314ecca"
>  
>  EXTRA_OECONF_class-native = "--without-gmp"
>  EXTRA_OECONF_class-target = "--enable-install-program=arch --libexecdir=${libdir}"
> @@ -53,6 +56,8 @@ sbindir_progs= "chroot"
>  # directory.
>  acpaths = "-I ./m4"
>  
> +FILES_${PN}-doc += "${mandir}"
> +
>  # Deal with a separate builddir failure if src doesn't exist when creating version.c/version.h
>  do_compile_prepend () {
>  	mkdir -p ${B}/src
> @@ -71,6 +76,10 @@ do_install_append() {
>  	# in update-alternatives to fail, therefore use lbracket - the name used
>  	# for the actual source file.
>  	mv ${D}${bindir}/[ ${D}${bindir}/lbracket.${BPN}
> +
> +	# prebuilt man pages
> +	install -d ${D}/${mandir}/man1
> +	install -t ${D}/${mandir}/man1 ${S}/man/*.1
>  }

One suggestion, in order for the man pages to match the "alternative names", and
also avoid conflicts with other potential suppliers.. I'd suggest enablign the
update-alternatives approach for the man pages as well.  Also rename the '[.1'
to 'lbracket.1.${BPN}' as part of that.

But the approach is correct, and I'm fine with this as-is if the community isn't
worried about man page clashes..

--Mark

>  do_install_append_class-native(){
>