[oe] [Angstrom-devel] [RFC] Per-ANGSTROM_MODE caches for Angstrom

[Richard Purdie]

On Wed, 2008-02-20 at 13:11 +0200, Paul Sokolovsky wrote:
> So, no objections? I'll add it then, and RFC adding to stable branch
> too. I'd like to keep that extra check for ANGSTROM_MODE being set - it
> at least consistent with existing MACHINE check.

See below.

> On Sat, 16 Feb 2008 17:52:42 +0200
> Paul Sokolovsky <pmiscml at gmail.com> wrote:
> 
> > > > --- conf/distro/include/angstrom.inc
> > > > b86b952c90304d1fd39f1415f6af7f71d02e00df +++
> > > > conf/distro/include/angstrom.inc
> > > > e7921869d30b7e5df3fd39fbd4bec5d2495b5d31 @@ -12,6 +12,7 @@
> > > > ANGSTROM_MODE ?= "glibc" # Can be "glibc", "eglibc" or "uclibc"
> > > > ANGSTROM_MODE ?= "glibc"
> > > >
> > > > +CACHE = "${TMPDIR}/cache${@['', '/' +
> > > > str(bb.data.getVar('ANGSTROM_MODE', d,
> > > > 1))][bool(bb.data.getVar('ANGSTROM_MODE', d, 1))]}${@['', '/' +
> > > > str(bb.data.getVar('MACHINE', d,
> > > > 1))][bool(bb.data.getVar('MACHINE', d, 1))]}" DEPLOY_DIR =
> > > > "${TMPDIR}/deploy/${ANGSTROM_MODE}" require
> > > > conf/distro/include/angstrom-${ANGSTROM_MODE}.inc
> > > 
> > > +1, for the  idea, but why not just
> > > ${TMPDIR}/cache/${ANGSTROM_MODE} ?
> > 
> > Well, nice question ;-). To be generic and putable even in
> > bitbake.conf ;-)

Please use the latter form, not the one with all the python in it. You
also need MACHINE in there.

> > > And I don't understand why you need to include MACHINE there.
> > > I've never needed to rebuild caches when switching MACHINE, and I
> > > shouldn't have to.
> > 
> > Heh, multimachine puts it there ;-)

In order to set MACHINE on the commandline, you need MACHINE in CACHE
and bitbake.conf does that for exactly this reason.

Cheers,

Richard