HEADS UP: New bsd.*.mk changes

[Marius Strobl]

On Tue, Jan 20, 2004 at 02:03:11PM -0500, Joe Marcus Clarke wrote:
> On Tue, 2004-01-20 at 13:41, Marius Strobl wrote:
> > On Tue, Jan 20, 2004 at 01:05:17PM -0500, Joe Marcus Clarke wrote:
> > > On Tue, 2004-01-20 at 13:02, Marius Strobl wrote:
> > > > On Tue, Jan 20, 2004 at 12:29:55PM -0500, Joe Marcus Clarke wrote:
> > > > > 
> > > > > I agree.  This approach seems the most flexible.  As for not being able
> > > > > to do non-root installs, this is a bogus argument as one could simply
> > > > > override PORT_DBDIR as they would PKG_DBDIR (even with the original
> > > > > patch).
> > > > > 
> > > > 
> > > > I was taking about non-root builds, e.g. single ports checked out
> > > > outside of PKGBASE to do maintenance work, not non-root installs. In
> > > > an environment where non-root installs are done your argument is valid.
> > > > Not being able to do non-root builds compilcates the job of maintainers.
> > > 
> > > As I said, non-root builds could override PORT_DBDIR the same way you
> > > could override PKG_DBDIR.  You could point that to a directory to which
> > > you could write.  What am I missing?
> > > 
> > 
> > In general I'd like to further on be able to build a port as non-root
> > and install als root, with default PREFIX, PKG_DBDIR and PORT_DBDIR.
> 
> Ah, I follow you.
> 
> > I don't see how this currently should work, if I set PORT_DBDIR to a
> > directory I can write to before I build the port as non-root it won't
> > neither read an existing OPTIONSFILE in the default location nor write
> > a possibly changed OPTIONSFILE to the new location. I could move around
> > the OPTIONSFILE before and after installing the port but that's really
> > messy.
> > At a first glance I don't see a reason why creation of PORT_DBDIR if
> > not already existing and writing of the OPTIONSFILE can't be done in
> > e.g. the fake-pkg target or a new target that's executed directly
> > before or after fake-pkg.
> 
> This may work, but it would require some reworking of the existing
> architecture, and, depending on patch complexity, may require another
> test build on bento.  Did you have something specific to review?
> 

No. I will look at it when the issue regarding the default for the
OPTIONSFILE and the layout of PORT_DBDIR, which currently is much
more important, has settled.