staging-tree: to stage or not to stage

staging-tree: to stage or not to stage

[Pavan Savoy]

Hi Greg,

<long mail warning...>

We've been having this discussion here and there, and there has been
various opinions about what staging is,
I've also read your mail @ http://lwn.net/Articles/285594/ on what the
tree IS and what it is NOT.

So, hoping you dispel all opinions and lay down some facts!!!

I think there is an opinion floating around that drivers in staging
lack some functionality or
don't work as expected or is just plain bad. (which might be true for
some drivers, but would be hard
on lot of people/drivers to generalize, I doubt any developer posting
to lkml would be irresponsible
enough to post an un-tested code...)

The reason I would stage any driver (if it is a self-containable
device driver) for that fact that a lot
of drivers get posted on lkml or other related sub-system list and
they keep on getting reviewed,
multiple versions come and go, and sometimes it so happens that all
the review comments are taken
care of, and MAINTAINER only comes back to say he wouldn't accept the
driver because he doesn't LIKE
it or thinks the underlying/related drivers is not good enough ....

So, why not stage everything, may be except few drivers which are
written by the MAINTAINERs themselves,
put up a TODO, get MAINTAINERs to review TODO, and get them to common
understanding that once the items in TODO are completed, it would be
in mainline...
As of today If I get 5 comments on my driver, I am not sure fixing
those 5 comments actually gets it to
mainline :(

regards,
Pavan

Re: staging-tree: to stage or not to stage

[Greg KH]

On Thu, Dec 23, 2010 at 04:11:43PM +0530, Pavan Savoy wrote:
> Hi Greg,
> 
> <long mail warning...>
> 
> We've been having this discussion here and there, and there has been
> various opinions about what staging is,
> I've also read your mail @ http://lwn.net/Articles/285594/ on what the
> tree IS and what it is NOT.
> 
> So, hoping you dispel all opinions and lay down some facts!!!

Sure, please ask specific questions and I will be glad to answer them.

I understand you are frustrated that some of your code had to be
rewritten numerous times, but that is just the nature of Linux kernel
development sorry.  I'm pretty sure that you agree that the end result
is much better than the original submissions, right?

thanks,

greg k-h