Dave,
> From an installation point of view, we don't currently care much - the
> projects need to arrive at a packaging breakdown which makes sense for
> them. We may choose to bring forward some recommendations or policies
> on package granularity based on the install strategy work I've been
> doing and the package/patch strategy work that's also getting under way,
> but what those might be is speculation at this point and I wouldn't try
> to impose them until they've been written down and reviewed. I don't
> think anything you're suggesting is unreasonable at first glance, though
ok, fine. I suppose, based on this I can try to go forward either with
Mike Kupfer or with another sponsor once I'm ready with the actual
changes.
> I wonder if some of the headers should just drop into SUNWhea rather
> than creating new packages for them.
The problem is that they are missing even from the closed tarballs, so
that's not an option if one wants to allow building SUNWhea even for a
non-closed build. Maybe those omissions are just a mistake and not
required for legal reasons. If so, that would certainly simplify things.
> The movement of files from one package to another between Solaris
> releases is handled by the package history mechanism, which is a set of
> metadata that's created during the product build and placed on the
> Solaris media at $CDROOT/Solaris_<version>/Product/.pkghistory for the
> installer to consume. Suffice to say, the changes you're suggesting can
> be dealt with easily, as far as I can see.
Great. I suppose there will be more visibility into this stuff once the
admin/install consolidation is opened?
> I don't know that this community can particularly guide you on the
> acceptability of specific solutions, though, as they cut across multiple
> technologies in the OS/Net consolidation, owners of which may or may not
> be represented on this list. You might do better to have this
> discussion on opensolaris-code instead, though we can certainly continue
> here if anyone has opinions or ideas to offer.
Ok. Since the changes seem reasonable to you and follow the same pattern
already used for several other drivers etc. in O/N, I'll either go ahead
and implement them as proposed or try to get another opinion on
opensolaris-code first.
Thanks.
Rainer
-----------------------------------------------------------------------------
Rainer Orth, Faculty of Technology, Bielefeld University
