On Wed, 20 Jun 2018 22:09:21 -0500
"Marty E. Plummer" <hanet...@startmail.com> wrote:
> On Thu, Jun 21, 2018 at 03:41:02AM +0100, M. J. Everitt wrote:
> > On 21/06/18 03:38, Jason Zaman wrote:
> > > On Wed, Jun 20, 2018 at 06:01:10PM -0500, Marty E. Plummer
> > > wrote:
> > >> On Wed, Jun 20, 2018 at 11:33:53PM +0100, James Le Cuirot
> > >> wrote:
> > >>> On Wed, 20 Jun 2018 17:21:09 -0500
> > >>> "Marty E. Plummer" <hanet...@startmail.com> wrote:
> > >>>
> > >>>> On Wed, Jun 20, 2018 at 09:03:44PM +0800, Jason Zaman wrote:
> > >>>>> On Wed, Jun 20, 2018 at 02:10:50AM -0500, Marty E. Plummer
> > >>>>> wrote:
> > >>>>>> Use ${EROOT%/} whereever possible, as the tools and
> > >>>>>> directories which are used with it are already prefixed with
> > >>>>>> a /
> > >>>>>>
> > >>>>>> Package-Manager: Portage-2.3.40, Repoman-2.3.9
> > >>>>>> ---
> > >>>>>> eclass/xdg-utils.eclass | 10 +++++-----
> > >>>>>> 1 file changed, 5 insertions(+), 5 deletions(-)
> > >>>>>>
> > >>>>>> diff --git a/eclass/xdg-utils.eclass
> > >>>>>> b/eclass/xdg-utils.eclass index ac075185d8e..8dba5ed6861
> > >>>>>> 100644 --- a/eclass/xdg-utils.eclass
> > >>>>>> +++ b/eclass/xdg-utils.eclass
> > >>>>>> @@ -66,7 +66,7 @@ xdg_environment_reset() {
> > >>>>>> # Updates the .desktop files database.
> > >>>>>> # Generates a list of mimetypes linked to applications that
> > >>>>>> can handle them xdg_desktop_database_update() {
> > >>>>>> - local
> > >>>>>> updater="${EROOT}${DESKTOP_DATABASE_UPDATE_BIN}"
> > >>>>>> + local
> > >>>>>> updater="${EROOT%/}${DESKTOP_DATABASE_UPDATE_BIN}"
> > >>>>> Shouldn't things like this be $BROOT since they're being run?
> > >>>>> $EROOT might be a different architecture that may or may not
> > >>>>> run at all on the build machine.
> > >>>>>
> > >>>> Good point, but here's a question; if
> > >>>> EROOT=${ROOT%/}${EPREFIX}, how do we use BROOT here? EBROOT?
> > >>>> Or longhand ${BROOT%/}${EPREFIX} ? I think that may be a use
> > >>>> case that got missed in the EAPI 7 discussions.
> > >>> BROOT is already prefixed as BROOT without a prefix would just
> > >>> be /.
> > >> I don't follow. Its my understanding that BROOT ~= ROOT for most
> > >> situations. But consider this setup:
> > >> Ubuntu amd64 with Gentoo Prefix, emerging a native arm @system to
> > >> /mnt/arm EPREFIX = /home/user/gentoo.
> > >>
> > >> In this situation, ROOT=/mnt/arm, EROOT=/mnt/arm, but what is
> > >> BROOT? /, or /home/usr/gentoo?
> > > https://dev.gentoo.org/~mgorny/articles/the-ultimate-guide-to-eapi-7.html#broot-variable-for-bdepend
> > >
> > > Basically BROOT already contains EPREFIX or BPREFIX or whatever
> > > it would be called. There is like no need for an un-prefixed
> > > BROOT so its just merged in. so you should just need
> > > "${BROOT}/usr/bin/update-mime-database"
> > >
> > Obligatory n00b question .. how does this work in EAPI <= 6 ?! :D
> >
> I would guess something like has eapi 7 || ROOT = BROOT or whatever.
> Use BROOT by default and if the EAPI doesn't support it set ROOT to
> BROOT or somat.
There was no variable for BROOT before EAPI 7, that's why we created
one! There was an internal Portage variable called
PORTAGE_OVERRIDE_EPREFIX, which is basically what BROOT gets set to now
but you should not use this in an eclass. I guess the safest fallback
would be EPREFIX. This would be technically wrong for cross-prefix
builds but unlikely to cause a problem in practise. Don't do
${BROOT-${EPREFIX}} though because BROOT is usually empty anyway.
--
James Le Cuirot (chewi)
Gentoo Linux Developer
