[oe] [oe-commits] Philip Balister : i2c-tools : Stage i2c-dev.h header as i2c-dev-user.h.

[Martin Jansa]

On Tue, Dec 14, 2010 at 08:30:46AM -0500, Philip Balister wrote:
> On 12/14/2010 07:06 AM, Dr. Michael Lauer wrote:
> > Am 10.12.2010 um 15:36 schrieb git version control:
> >
> >> Module: openembedded.git
> >> Branch: org.openembedded.dev
> >> Commit: 48e6a063370a38a35f31a28efd8f6ce6ebf00840
> >> URL:    http://gitweb.openembedded.net/?p=openembedded.git&a=commit;h=48e6a063370a38a35f31a28efd8f6ce6ebf00840
> >>
> >> Author: Philip Balister<philip at balister.org>
> >> Date:   Fri Dec 10 09:31:01 2010 -0500
> >>
> >> i2c-tools : Stage i2c-dev.h header as i2c-dev-user.h.
> >>
> >> Thanks to John Faith for suggesting this approach on the ML. The problem is
> >> i2c-tools overwrites the header staged by the kernel. This breaks programs
> >> that depend on the kernel header.
> >
> > I don't think this a good solution. Now all programs break which expect
> > this very i2c-dev.h as being staged by i2c-tools. Interestingly, the
> > desktop distros don't bother about staging it differently, so why
> > can't we do the same? Besides, the i2c-dev.h as staged by i2c-tools
> > is supposed to be a superset. If it isn't we should complain with the
> > i2c-tools developers.
> >
> > What can we do in the meantime?
> 
> Can we get a survey of how other distros handle the situation?

Gentoo
i2c-tools-3.0.[012] (all versions)
src_install() {
    emake install prefix="${D}"/usr || die
    rm -rf "${D}"/usr/include # part of linux-headers
...
}

so it's not part of i2c-tools package (gentoo doesn't have concept of
-dev packages as it's always usefull to have all -dev stuff on source
based distribution).

-- 
Martin 'JaMa' Jansa     jabber: Martin.Jansa at gmail.com