Re: [gentoo-python] Re: [PATCH updated 1/3] Set env for best Python impl in *_all() phases.

[Michał Górny]

On Mon, 03 Dec 2012 10:46:42 -0500
Mike Gilbert <floppym@g.o> wrote:

> On 12/3/2012 6:00 AM, Michał Górny wrote:
> > Export the best implementation's build dir as BEST_BUILD_DIR, in order
> > to avoid clobbering original BUILD_DIR.
> > ---
> >  gx86/eclass/distutils-r1.eclass | 31 ++++++++++++++++++++++++++-----
> >  1 file changed, 26 insertions(+), 5 deletions(-)
> >
> > diff --git a/gx86/eclass/distutils-r1.eclass b/gx86/eclass/distutils-r1.eclass
> > index b1b3f90..db8fe86 100644
> > --- a/gx86/eclass/distutils-r1.eclass
> > +++ b/gx86/eclass/distutils-r1.eclass
> > @@ -388,6 +388,27 @@ distutils-r1_run_phase() {
> >  	if [[ ${DISTUTILS_IN_SOURCE_BUILD} ]]; then
> >  		popd &>/dev/null || die
> >  	fi
> > +
> > +	# Store them for reuse.
> > +	_DISTUTILS_BEST_IMPL=(
> > +		"${EPYTHON}" "${PYTHON}" "${BUILD_DIR}" "${PYTHONPATH}"
> > +	)
>
> Why are you using an array here? To save 1 line of code?

Yes, something like that. Plus do not have to invent names for all
of them.

> Putting the values in an array makes them anonymous for no reason I can see.

To be honest, I don't see a reason for them to be non-anonymous. It's
not something user should touch. A bit like four variables == four
times as much risk of random collision.

> > +}
> > +
> > +# @FUNCTION: _distutils-r1_run_common_phase
> > +# @USAGE: [<argv>...]
> > +# @INTERNAL
> > +# @DESCRIPTION:
> > +# Run the given command, restoring the best-implementation state.
> > +_distutils-r1_run_common_phase() {
> > +	local EPYTHON=${_DISTUTILS_BEST_IMPL[0]}
> > +	local PYTHON=${_DISTUTILS_BEST_IMPL[1]}
> > +	local BEST_BUILD_DIR=${_DISTUTILS_BEST_IMPL[2]}
> > +	local PYTHONPATH=${_DISTUTILS_BEST_IMPL[3]}
> > +
> > +	export EPYTHON PYTHON PYTHONPATH
> > +
> > +	"${@}"
> >  }
> >
> >  distutils-r1_src_prepare() {
> > @@ -419,7 +440,7 @@ distutils-r1_src_configure() {
> >  	multijob_finish
> >
> >  	if declare -f python_configure_all >/dev/null; then
> > -		python_configure_all
> > +		_distutils-r1_run_common_phase python_configure_all
> >  	fi
> >  }
> >
> > @@ -435,7 +456,7 @@ distutils-r1_src_compile() {
> >  	multijob_finish
> >
> >  	if declare -f python_compile_all >/dev/null; then
> > -		python_compile_all
> > +		_distutils-r1_run_common_phase python_compile_all
> >  	fi
> >  }
> >
> > @@ -451,7 +472,7 @@ distutils-r1_src_test() {
> >  	multijob_finish
> >
> >  	if declare -f python_test_all >/dev/null; then
> > -		python_test_all
> > +		_distutils-r1_run_common_phase python_test_all
> >  	fi
> >  }
> >
> > @@ -467,9 +488,9 @@ distutils-r1_src_install() {
> >  	multijob_finish
> >
> >  	if declare -f python_install_all >/dev/null; then
> > -		python_install_all
> > +		_distutils-r1_run_common_phase python_install_all
> >  	else
> > -		distutils-r1_python_install_all
> > +		_distutils-r1_run_common_phase distutils-r1_python_install_all
> >  	fi
> >  }
> >
>
> I haven't run into a situation that patch series is intended to address,
> so I don't have any strong opinion on it. If you think it is useful,
> fine by me.

Well, the common case I can think of is building docs. Although it
isn't really necessary to bind that strongly to any specific
implementation, doing so allows packages to use compiled modules.

-- 
Best regards,
Michał Górny