On 2008/09/05 15:20, Marc Espie wrote:
> On Fri, Sep 05, 2008 at 12:38:15PM +0100, Stuart Henderson wrote:
> > > COMMENT = Rockbox Utility
> >
> > this could be a bit more descriptive, maybe something like
> > "tool to install Rockbox firmware on portable music players"
> >
> Too long
> install rockbox firmware on ipod
>
> (does it apply to other mp3 players ?)
yes, they have code for old ipods and some other players.
they haven't reverse-engineered new ipods (classic/etc) far
enough yet.
> > > WRKBUILD = ${WRKDIR}/rockbox-*/rbutil/rbutilqt
> > > WRKDIST = ${WRKDIR}/rockbox-*/rbutil
>
> NO WILDCARD in WRKBUILD and WRKDIST, this does *not* work, as soon as
> you update the port, if anyone has got a wrkdir checked out, this will
> *fail* in a momentous way.
indeed. I think this is what was breaking update-patches before
too, I fixed it in the tarball I sent but forgot to comment on it.
there is actually a related problem which happens sometimes with
ports with subpackages: if a patch is applied which affects only
one subpackage, so only that subpackage's version gets bumped,
you have to make sure WRKDIR is clean. I've seen this from time
to time, don't know if there's anything currently doing it.
In that instance, should PKGNAME get bumped every time any
subpackage is bumped?
i.e. starting with this contrived example
PKGNAME=${DISTNAME}
PKGNAME-main=${DISTNAME}
PKGNAME-foo=${DISTNAME}-foo
moving to this ...
PKGNAME=${DISTNAME}
PKGNAME-main=${DISTNAME}
PKGNAME-foo=${DISTNAME}p0-foo
requires cleaning WRKDIR otherwise if the WRKDIR exists already,
it won't pick up the update correctly (wrong patches/whatever).
but if instead we do this, it's avoided:-
PKGNAME=${DISTNAME}p0
PKGNAME-main=${DISTNAME}
PKGNAME-foo=${DISTNAME}p0-foo
I don't think this gets done very often, if at all, but maybe
we should start.. thoughts?
