On 2016/03/08 16:41, Nigel Taylor wrote:
> modbuild shouldn't be used but maybe a modtiny style adding an extra
> style doesn't impact existing modbuild ports.

My build is still running, but no problems to report yet. If nothing
shows up as a problem I don't see why we couldn't use the same
"MODPERL_configure" for both. And actually we could do something
like this, it would be quite intuitive: set CONFIGURE_STYLE to
either just "modbuild" or to "modbuild tiny".

.if ${CONFIGURE_STYLE:L:Mmodbuild}
.  if ${CONFIGURE_STYLE:L:Mtiny}
BUILD_DEPENDS +=        devel/p5-Module-Build-Tiny
.  else
BUILD_DEPENDS +=        devel/p5-Module-Build
.  fi
MODPERL_configure = \
        arch=`perl -e 'use Config; print $$Config{archname}, "\n";'`; \
    cd ${WRKSRC}; ${_SYSTRACE_CMD} ${SETENV} ${CONFIGURE_ENV} \
        perl Build.PL \
                --install_path=lib="${PREFIX}/libdata/perl5/site_perl" \
                --install_path=arch="${PREFIX}/libdata/perl5/site_perl/$$arch" \
                --install_path=libdoc="${PREFIX}/man/man3p" \
                --install_path=bindoc="${PREFIX}/man/man1" \
                --install_path=bin="${PREFIX}/bin" \
                --install_path=script="${PREFIX}/bin" ${CONFIGURE_ARGS}
.else
[...]

If ::Tiny diverges further in future we could do something different
but this seems a cleaner approach for now.

> diff -u -p -r1.12 Makefile
> --- Makefile    12 Jun 2015 13:46:08 -0000      1.12
> +++ Makefile    8 Mar 2016 15:12:07 -0000
> @@ -17,9 +17,10 @@ RUN_DEPENDS =        devel/p5-Carp-Clan \
>                 devel/p5-Sub-Name \
>                 devel/p5-namespace-autoclean
>  BUILD_DEPENDS = ${RUN_DEPENDS} \
> -               devel/p5-Module-Build-Tiny
> 

this leaves a dangerous trailing \

>  .  if ${CONFIGURE_STYLE:L:Mmodinst}
>  BUILD_DEPENDS +=     devel/p5-Module-Install
>  CONFIGURE_ARGS +=    --skipdeps
> -_MODPERL_preconfig = rm -rf ${WRKSRC}/inc/Module/*Install*
> +_MODPERL_preconfig = rm -rf ${WRKSRC}/inc/*
>  .  else
>  _MODPERL_preconfig = :
>  .  endif

this looks like part of a different diff

Reply via email to