[PATCH] MAINTAINERS: Start using the 'reviewer' (R) tag

[Bartlomiej Zolnierkiewicz]

[ this time with full Cc: & context preserved ]

Hi,

On Wednesday, October 28, 2015 08:24:46 AM Lee Jones wrote:
> On Tue, 2015-10-27 at 18:15 +0000, Lee Jones wrote:
> > On Tue, 27 Oct 2015, Sebastian Reichel wrote:
> > > On Tue, Oct 27, 2015 at 03:42:37PM +0000, Lee Jones wrote:
> > > > Since eafbaac ("MAINTAINERS: Add "R:" designated-reviewers tag") we
> > > > have been able to tag specific people as Reviewers.  These are key
> > > > individuals who are tasked with or volunteer to review code submitted
> > > > to a subsystem or specific file.  However, according to MAINTAINERS
> > > > we have 1046 Maintainers and only a mere 22 Reviewers.  I believe
> > > > these numbers to be incorrect, as many of these Maintainers are in
> > > > fact Reviewers.
> 
> Most entries in MAINTAINERS seem to be vanity entries than actual
> active participants.  A person typically writes a driver, adds a
> MAINTAINER entry, then forgets about it and/or the hardware becomes
> outdated.
> 
> This I agree with.
> 
> On Wed, 28 Oct 2015, Krzysztof Kozlowski wrote:
> > 2015-10-28 3:44 GMT+09:00 Joe Perches <joe at perches.com>:
> > > On Tue, 2015-10-27 at 18:15 +0000, Lee Jones wrote:
> > > > On Tue, 27 Oct 2015, Sebastian Reichel wrote:> >
> > > > > I think you should CC the people, which are changed from "M:" to
> > > > > "R:", though.
> > > >
> > > > Yes, makes sense.
> > > >
> > > > I'd like to collect some Maintainer Acks first though.
> > >
> > > I think people from organizations like Samsung are actual
> > > maintainers not reviewers.
> 
> So this all hinges on how we are describing Maintainers and Reviewers.
> 
> My personal definition (until convinced otherwise) is that Reviewers
> care about their particular subsystem and/or files.  They conduct code
> reviews to ensure nothing gets broken and the code base stays in best
> possible state of worthiness.  On the other hand Maintainers usually
> conduct themselves as Reviewers but also have 'maintainership' duties
> as well; such as applying patches, *maintaining*, testing, rebasing,
> etc, an upstream branch and ultimately sending pull-requests to higher
> level Maintainers i.e. Linus.  Maintainers also have the ultimate say
> (unless over-ruled by Linus etc) over what gets applied.
> 
> > > Their drivers are not thrown over a wall and forgotten.
> > 
> > At least for Samsung Multifunction PMIC drivers (and some of Maxim
> > MUICs and PMICs) these are actively used by us in existing and new
> > products. They are also continuously extended and actually maintained.
> > This means that it is not only about review of new patches but also
> > about caring that nothing will become broken.
> 
> Exactly.  This what I expect of any good code Reviewer.
> 
> > I would prefer to leave the "SAMSUNG MULTIFUNCTION PMIC DEVICE
> > DRIVERS" entry as is - maintainers.
> 
> But you aren't maintaining the driver i.e. you don't collect patches
> and *maintain* them on an upstream branch.  Granted some of you guys
> are doing a great job of maintaining branches on your downstream or
> BSP kernels, but conduct a Reviewer type role for upstream.
> 
> You guys are pushing back like this is some kind of demotion.  That's
> not the case at all.  All it does is better describe the (very worthy)
> function you *actually* provide.

It is actually a demotion from my POV:

* "Reviewer" doesn't accurately describe the job of doing all the needed
  testing, bug-fixing and additional contributions that is often done by
  people without their own branches.

* You don't know internal policies of all companies involved in Linux
  Kernel development.  "Maintainer" is a well known term and sometimes
  person's job status or "key performance indicators" may depend on it.

Best regards,
--
Bartlomiej Zolnierkiewicz
Samsung R&D Institute Poland
Samsung Electronics