On Monday, 17 December 2007, at 12:55:58 (-0500),
Jeff Johnson wrote:

> (aside) Which means everything is in place to remove the "standard"
> packaging boilerplate at the beginning of the %install build
> scriptlet that tries to trash-and-burn any pre-existing
> %{buildroot}. Not that anyone is going to be brave enough to permit
> rpm to automate %{buildroot} handling, judging from past experience,
> it will be another decade or more before any vendor deviates from
> what they have always done. That ain't my rpm problem any more ...
>
> OTOH, the change is highly likely to break odd-ball packaging like
> what is usually found in vendor kernel packages, which typically are
> working at cross-purposes with rpmbuild anyways, attempting to loop
> over multiple builds for arches (and xen and ...) with some weird
> %buildroot trickery.
>
> Note that %_spec_install_pre is just a macro, you can redefine to
> %{nil} and do whatever you please, in kernel.spec.

I think the numerous discussions and examples we've posted to various
lists about how to redefine that behavior if needed should more than
suffice.  Feel free to pass any complainers off to me, but the sample
spec files are out there for anyone with sufficient congitive
capability to look for them.

Personally I don't build packages with that extra fluff in there
anyway since Mezz has always handled supplying clean buildroots to
package builds.  But I do see the construct a lot.

> Or just blame me for introducing "incompatibilities" and "useless
> features" in rpm-5.0, that works just fine too.

That goes without saying.  :-)

Michael

-- 
Michael Jennings (a.k.a. KainX)  http://www.kainx.org/  <mej@kainx.org>
Linux Server/Cluster Admin, LBL.gov       Author, Eterm (www.eterm.org)
-----------------------------------------------------------------------
 "Persistence trumps talent and looks every time."      -- Aaron Brown