Re: [gentoo-dev] Re: [gentoo-commits] gentoo-x86 commit in dev-python/pygobject: ChangeLog pygobject-2.14.0.ebuild

[Mart Raudsepp]

On N, 2007-09-27 at 17:08 +0100, Steve Long wrote:
> Donnie Berkholz wrote:
>
> > On 13:15 Wed 26 Sep     , Mart Raudsepp (leio) wrote:
> >> mv "${D}"/usr/$(get_libdir)/python${PYVER}/site-packages/pygtk.py \
> >> "${D}"/usr/$(get_libdir)/python${PYVER}/site-packages/pygtk.py-2.0
> >> mv "${D}"/usr/$(get_libdir)/python${PYVER}/site-packages/pygtk.pth \
> >> "${D}"/usr/$(get_libdir)/python${PYVER}/site-packages/pygtk.pth-2.0
> >> }
> >>
> >> pkg_postinst() {
> >> python_version
> >> python_mod_optimize /usr/lib/python${PYVER}/site-packages/gtk-2.0
> >> alternatives_auto_makesym
> >> /usr/$(get_libdir)/python${PYVER}/site-packages/pygtk.py
> >> pygtk.py-[0-9].[0-9] alternatives_auto_makesym
> >> /usr/$(get_libdir)/python${PYVER}/site-packages/pygtk.pth
> >> pygtk.pth-[0-9].[0-9] python_mod_compile
> >> ${ROOT}usr/$(get_libdir)/python${PYVER}/site-packages/pygtk.py
> >
> > This is pretty weird, because it uses get_libdir() almost everywhere but
> > skips it once. Wonder if we could add a repoman check that looks for
> > whether an ebuild calls get_libdir(), then checks for any instances of
> > '[^[:alnum:]]lib[^[:alnum:]]'.
> >
> It seems a bit inefficient to keep calling the function in an implicit
> subshell. What's wrong with ldir="/usr/$(get_libdir)/" [or "${ROOT}usr.."]
> once and then reusing the value?

Readability in my eyes. This function call can't cost more than all the
disk activity that is going to go on from the calls..

> (The single use of $ROOT in pkg_postinst() seems odd too, to this newb.)

That's what python.eclass gives you. Other functions takes care of
${ROOT} for you, but that doesn't.


As a side note, the python_mod_optimize lib was changed to a
$(get_libdir) yesterday based on Donnies initial post here.

-- 
Mart Raudsepp
Gentoo Developer
Mail: leio@g.o
Weblog: http://planet.gentoo.org/developers/leio