Re: [PATCH 1/3] perl: fix Perl5 module builds

[Jens Rehsack <rehsack@gmail.com>]

> Am 08.10.2015 um 16:21 schrieb Jens Rehsack <rehsack@gmail.com>:
> 
> This patch fixes some issues in classes providing cpan module build support:
> 
> * add support even for xs modules with more than 3 levels as
>  B::Hooks::End::Of::Scope or Math::Random::ISAAC::XS
> * correct handling of Module::Build (as far as stolen from pkgsrc
>  and my humble knowledge)
> * configure to install to vendor_libs as default, even when
>  inherited do_install remains unused (overwritten do_install)
> 
> Signed-off-by: Jens Rehsack <sno@netbsd.org>
> ---
> meta/classes/cpan-base.bbclass  |  4 ++++
> meta/classes/cpan.bbclass       |  2 +-
> meta/classes/cpan_build.bbclass | 13 +++++++++++--
> 3 files changed, 16 insertions(+), 3 deletions(-)
> 
> diff --git a/meta/classes/cpan-base.bbclass b/meta/classes/cpan-base.bbclass
> index d9817ba..7810a4d 100644
> --- a/meta/classes/cpan-base.bbclass
> +++ b/meta/classes/cpan-base.bbclass
> @@ -49,7 +49,11 @@ PERLVERSION[vardepvalue] = ""
> FILES_${PN}-dbg += "${PERLLIBDIRS}/auto/*/.debug \
>                     ${PERLLIBDIRS}/auto/*/*/.debug \
>                     ${PERLLIBDIRS}/auto/*/*/*/.debug \
> +                    ${PERLLIBDIRS}/auto/*/*/*/*/.debug \
> +                    ${PERLLIBDIRS}/auto/*/*/*/*/*/.debug \
>                     ${PERLLIBDIRS}/vendor_perl/${PERLVERSION}/auto/*/.debug \
>                     ${PERLLIBDIRS}/vendor_perl/${PERLVERSION}/auto/*/*/.debug \
>                     ${PERLLIBDIRS}/vendor_perl/${PERLVERSION}/auto/*/*/*/.debug \
> +                    ${PERLLIBDIRS}/vendor_perl/${PERLVERSION}/auto/*/*/*/*/.debug \
> +                    ${PERLLIBDIRS}/vendor_perl/${PERLVERSION}/auto/*/*/*/*/*/.debug \
>                     "
> diff --git a/meta/classes/cpan.bbclass b/meta/classes/cpan.bbclass
> index e2bbd2f..8e079e0 100644
> --- a/meta/classes/cpan.bbclass
> +++ b/meta/classes/cpan.bbclass
> @@ -17,7 +17,7 @@ export PERLHOSTLIB = "${STAGING_LIBDIR_NATIVE}/perl-native/perl/${@get_perl_vers
> 
> cpan_do_configure () {
> 	export PERL5LIB="${PERL_ARCHLIB}"
> -	yes '' | perl ${EXTRA_PERLFLAGS} Makefile.PL ${EXTRA_CPANFLAGS}
> +	yes '' | perl ${EXTRA_PERLFLAGS} Makefile.PL INSTALLDIRS=vendor ${EXTRA_CPANFLAGS}
> 
> 	# Makefile.PLs can exit with success without generating a
> 	# Makefile, e.g. in cases of missing configure time
> diff --git a/meta/classes/cpan_build.bbclass b/meta/classes/cpan_build.bbclass
> index 4f648a6..3653748 100644
> --- a/meta/classes/cpan_build.bbclass
> +++ b/meta/classes/cpan_build.bbclass
> @@ -8,6 +8,7 @@ EXTRA_CPAN_BUILD_FLAGS ?= ""
> # Env var which tells perl if it should use host (no) or target (yes) settings
> export PERLCONFIGTARGET = "${@is_target(d)}"
> export PERL_ARCHLIB = "${STAGING_LIBDIR}${PERL_OWN_DIR}/perl/${@get_perl_version(d)}"
> +export PERLHOSTLIB = "${STAGING_LIBDIR_NATIVE}/perl-native/perl/${@get_perl_version(d)}/"
> export LD = "${CCLD}"
> 
> cpan_build_do_configure () {
> @@ -24,14 +25,22 @@ cpan_build_do_configure () {
> 				--install_path bindoc=${mandir}/man1 \
> 				--install_path libdoc=${mandir}/man3 \
>                                 ${EXTRA_CPAN_BUILD_FLAGS}
> +
> +	# Build.PLs can exit with success without generating a
> +	# Build, e.g. in cases of missing configure time
> +	# dependencies. This is considered a best practice by
> +	# cpantesters.org. See:
> +	#  * http://wiki.cpantesters.org/wiki/CPANAuthorNotes
> +	#  * http://www.nntp.perl.org/group/perl.qa/2008/08/msg11236.html
> +	[ -e Build ] || bbfatal "No Build was generated by Build.PL"
> }
> 
> cpan_build_do_compile () {
> -        perl Build
> +        perl Build verbose=1
> }
> 
> cpan_build_do_install () {
> -	perl Build install
> +	perl Build install --destdir ${D}
> }
> 
> EXPORT_FUNCTIONS do_configure do_compile do_install
> -- 
> 2.3.8 (Apple Git-58)
> 
> -- 
> Jens Rehsack - rehsack@gmail.com

Since it's two weeks ago since submitted and this patch series fixes several problems how yocto deals with perl module distributions, I'd like to get a more concrete explanation why an infrastructure fix isn't validated during feature freeze for branching.

None of the three patches adds a feature, they all fix broken infrastructure of perl module handling in Yocto which should better now validated than after branching.
Since after freeze the master branch might become unstable for a while, I might not be able to re-submit.

Cheers
-- 
Jens Rehsack - rehsack@gmail.com