On Tue, Mar 11, 2014 at 8:41 AM, Matthew Miller
<mattdm(a)fedoraproject.org> wrote:
On Tue, Mar 11, 2014 at 08:19:38AM -0400, Josh Boyer wrote:
> So we're already smaller than Ubuntu. Size is one of the things we've
> been told is key to adoption in the cloud. If we're already smaller,
> and Ubuntu is more widely adopted, I don't see that holding true.
> What am I missing, or is the "we need a smaller kernel package" thing
> somehow trying to pull the wool over my eyes?
Oh, it absolutely remains nice to have, and by "nice" I mean
"awesome".
If we can improve all four of the big size consumers (python, kernel,
modularized docs, modularized i18n), it'll go from "on par" to actually
being an advantage. And even if we can't get to that for a while,
improvement is still improvement, and when I said size isn't the primary
driver, I certainly didn't mean to imply that it's not an important one.
Since isn't the only driver for the kernel work you're doing either, I don't
think. Another big advantage is that the modularized drivers will allow us
to skip out-of-schedule updates for security updates in the driver package.
Erm.. only if you manually pay attention to where the driver is for a
particular CVE. It's a subpackage, not a separate package. As far as
I know, the yum security metadata thing applies to packages as an
entire set, not each subpackage. I believe that means when we build a
kernel for a CVE fix, you're still going to see a kernel-core update
available being marked as a security fix.
josh