On Mon, Jul 22, 2013 at 5:49 PM, Matthew Miller
<mattdm(a)fedoraproject.org> wrote:
On Mon, Jul 22, 2013 at 05:39:33PM +0200, Miloslav Trmač wrote:
> This proposal doesn't seem to make rings 0 or 1 particularly more
> integrated - just smaller; and it places a higher barrier to entry to
> adding useful functionality to the core.
I didn't intend to say that.
To clarify (and ignoring the "higher barrier" that's discussed in the
second paragraph), are you actually proposing that ring 0/1 should be
more integrated than what we are doing now?
> Want to add new a JSON library, HTTPS library with OCSP support,
DNS
> resolver to the integrated core? Sorry, 30% of the stacks are on a
> 2-year lifecycle and can't use it now; other 30% have decided to ship
> unmodified upstream code and will not integrate any Fedora
> Core-specific patches.
I definitely didn't intend to say the "sorries" you've put there. In
fact,
quite the opposite. How can I make that more clear?
It seems to me that these are inevitable consequences of giving SIGs
more freedom and power to override some policies: if we allow SIGs to
avoid RPM, why wouldn't we allow them to avoid libmyperfectdns that
has been added to Ring 0? (And this is why I think the "higher
barrier" will arise in practice - we can make the core better all we
want, but getting the new features adopted will be more difficult than
now.)
Mirek