On 11/14/2016 12:36 PM, Paul Bolle wrote:
On Mon, 2016-11-14 at 14:47 -0500, Josh Boyer wrote:
> You keep saying things like that, but I honestly have lost all context
> of what you are actually wanting to change here and why.
My overall goals are pretty vague, so I suppose my objections are pretty vague
too. My goals for this series are pretty clear, however
And as far as this series is concerned: only patch 3/5 is controversial,
right?
> Could you start with a summary email of the workflow you are desiring,
> with some examples perhaps? That would be helpful.
Well, hmm. I guess there are three kind of reasons to build the kernel rpm
locally.
(If none of these reasons make sense, or, even worse, you suspect almost no
one is actually building the kernel rpm locally in the first place we might as
well stop this discussion right here. In that case I'm inclined to take "_we_
don't care" as a reasonable objection to any patches I submit.)
1) Rebuilding a minimal set of the kernel packages as quickly, basically, etc.
as possible. Probably to test the stack of local patches one has. That's what
I do. It requires "--with vanilla" and a few "--without" rpmbuild
flags.
2) Rebuilding the entire set of kernel packages for your local architecture,
with all Fedora specific patches, etc. Probably also to test some local
patches. I never do that.
3) Play Fauxdora and rebuild all kernel packages for all supported
architectures. I never do that, and can't imagine I'll ever want to do that.
Anyhow, my sort of, kind a goal is to make 1) as easy as possible. Does this
clarify things a bit?
These are all fine reasons and motivations but I'm not seeing
the benefit of this patch specifically. The cost to generate
a few extra arch configs is negligible compared to the actual
time spent building the kernel. Do you have numbers showing
how much your build time is sped up?
Paul Bolle
Thanks,
Laura