On Fri, Feb 03, 2017 at 12:25:24PM +0000, Emil Velikov wrote:

On 3 February 2017 at 08:00, Daniel Vetter [off-list ref] wrote:

quoted

On Thu, Feb 02, 2017 at 12:37:21PM +0000, Emil Velikov wrote:

quoted

 - Daniel, gents - is drm-misc aimed at devs with limited (no?)
review/commit history in the area and/or the kernel in general ?
In this case, Peter have quite noticeable experience [in kernel
development] with little-to no in DRM.
Should one draw a line in the sand somewhere ?

We're fairly lenient with accepting new drivers already, and for drivers
in drm-misc we require some checks and peer review to make sure new folks
learn faster. But it's an expirement, I fully expect some fallout and then
some learnign and fine tuning from that, and maybe we even need to crank
requirements back up to a much higher level of experience.

But for drivers I'm honestly not too concerned: As long as there's some
process in place to foster learning&collaboration (which I think will be
better with this) I really don't see much harm in merging non-perfect
driver code.

Thank you Daniel, this is very well said. Being lenient to begin with
and adjusting where/if needed is the more welcoming approach, indeed.
Doubt there'll be (m)any cases of (ab|mis)use but even if so, nobody
can see the future.

Yeah, the big idea behind drm-misc is also to make drm more welcoming for
new folks, and I think starting with a small driver or a few driver
patches for hw you have lying around for hacking is a great way. And every
once in a while we can "trick" one of these newbies into doing some cool
core refactoring or cleanup, and maybe even to stick around for a while.

All for world domination and other evil purposes ofc :-)
-Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch