On Thu 03 Dec 2009 at 09:54AM, Edward Pilatowicz wrote:
> On Thu, Dec 03, 2009 at 08:48:57AM -0700, Jerry Jelinek wrote:
> > Ed,
> >
> > I've updated the code to address your comments above.
> > As per my response yesterday, I've left the Makefile as is.
> > There is an updated webrev at:
> >
> > http://cr.opensolaris.org/~gjelinek/webrev.8878/
> >
> > Let me know if you have any other comments.
So this code seems to further propagate into more corners the
notion that everything we do revolves around 'entire'-- but
we know that's not right, and will break down pretty darn soon
when Liane puts back.
It seems to me that it'd be a lot better to let the manifest
carry along a blob of data which the packaging system gives
it-- perhaps inside of <[[CDATA, and then to give that
same blob back to the pkg system on the remote side.
Otherwise, it seems like a pkg FMRI is a somewhat incomplete
identifier, since we lack any sort of publisher information with which
to give it context. It will work most of the time, but there seems
to be a nice chance here to better define the architectural
responsibilities for interpreting different aspects of the manifest.
The current <patch>, <obsoletes>, <package>, etc, seem to me to
be an artifact of the old reality of an unchanging and unchangeable
SVR4 packaging system-- we taught zones about its particulars
because we had to. Finding a way to delete as much of that as possible
seems desirable.
-dp
--
Daniel Price, Solaris Kernel Engineering http://blogs.sun.com/dp
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
