Xen devel - Mar 2013 - [RFC] CCing MAINTAINERS on patches by default?

With the numbers of patches flying around it can be a bit hard for a
maintainer to spot patches which (s)he is expected to comment on / deal
with, especially now that we have a larger number of maintainers of
varying subsystems etc.

I think it might be time to start asking patch submitters to CC the
maintainers of the code they are touching, based on the MAINTAINERS
file.

If people (specifically maintainers) agree with this change then we can
start by updating the "Submitting Patches" docs on the wiki etc and
gently start requesting submitters to add the CCs.

In the ultimate irony there are too many maintainers for me to CC here,
so I''ve just gone with the committers ;-)

Ian.

At 17:00 +0000 on 12 Mar (1363107618), Ian Campbell
wrote:
> With the numbers of patches flying around it can be a bit hard for a
> maintainer to spot patches which (s)he is expected to comment on / deal
> with, especially now that we have a larger number of maintainers of
> varying subsystems etc.
> 
> I think it might be time to start asking patch submitters to CC the
> maintainers of the code they are touching, based on the MAINTAINERS
> file.
> 
> If people (specifically maintainers) agree with this change then we can
> start by updating the "Submitting Patches" docs on the wiki etc
and
> gently start requesting submitters to add the CCs.
Fine by me; I usually work from my xen-devel inbox but I don''t object
to
being Cc''d directly.

Can we provide git/hg runes to add Cc:s automatically?

Tim.

On 12/03/2013 17:00, "Ian Campbell" <Ian.Campbell@citrix.com>
wrote:
> With the numbers of patches flying around it can be a bit hard for a
> maintainer to spot patches which (s)he is expected to comment on / deal
> with, especially now that we have a larger number of maintainers of
> varying subsystems etc.
> 
> I think it might be time to start asking patch submitters to CC the
> maintainers of the code they are touching, based on the MAINTAINERS
> file.
> 
> If people (specifically maintainers) agree with this change then we can
> start by updating the "Submitting Patches" docs on the wiki etc
and
> gently start requesting submitters to add the CCs.
Fine by me!

 -- Keir
> In the ultimate irony there are too many maintainers for me to CC here,
> so I''ve just gone with the committers ;-)
> 
> Ian.
>

On Tue, 2013-03-12 at 17:07 +0000, Tim Deegan wrote:
> At 17:00 +0000 on 12 Mar (1363107618), Ian Campbell wrote:
> > With the numbers of patches flying around it can be a bit hard for a
> > maintainer to spot patches which (s)he is expected to comment on /
deal
> > with, especially now that we have a larger number of maintainers of
> > varying subsystems etc.
> > 
> > I think it might be time to start asking patch submitters to CC the
> > maintainers of the code they are touching, based on the MAINTAINERS
> > file.
> > 
> > If people (specifically maintainers) agree with this change then we
can
> > start by updating the "Submitting Patches" docs on the wiki
etc and
> > gently start requesting submitters to add the CCs.
> 
> Fine by me; I usually work from my xen-devel inbox but I don''t
object to
> being Cc''d directly.
> 
> Can we provide git/hg runes to add Cc:s automatically?
In git you just add "Cc: Someone <someone@exmaple@com>" in your
commit
next to the S-o-b and git send email does the right thing. AFAIK hg
doesn''t do anything similar.

Going further we could import Linux''s ./sciprts/get_maintainer.pl
(which
takes a patch, parses MAINTAINERS and outputs the result). Unfortunately
the Linux default is to also look at S-o-b and git blame, so it has a
tendency to spam everyone who ever went anywhere near the file. We
don''t
have enoguh subsystems/MAINTAINERS that I think this is a huge
requirement -- we could leave it for now and see how we get on IMHO

Ian.

Ian Campbell writes ("[RFC] CCing MAINTAINERS on patches by
default?"):
> If people (specifically maintainers) agree with this change then we can
> start by updating the "Submitting Patches" docs on the wiki etc
and
> gently start requesting submitters to add the CCs.
I think this is a good idea.

Ian.

On Tue, Mar 12, 2013 at 05:07:42PM +0000, Tim Deegan
wrote:
> At 17:00 +0000 on 12 Mar (1363107618), Ian Campbell wrote:
> > With the numbers of patches flying around it can be a bit hard for a
> > maintainer to spot patches which (s)he is expected to comment on /
deal
> > with, especially now that we have a larger number of maintainers of
> > varying subsystems etc.
> > 
> > I think it might be time to start asking patch submitters to CC the
> > maintainers of the code they are touching, based on the MAINTAINERS
> > file.
> > 
> > If people (specifically maintainers) agree with this change then we
can
> > start by updating the "Submitting Patches" docs on the wiki
etc and
> > gently start requesting submitters to add the CCs.
> 
> Fine by me; I usually work from my xen-devel inbox but I don''t
object to
> being Cc''d directly.
> 
> Can we provide git/hg runes to add Cc:s automatically?
That would be neat. Linux uses scripts/get_maintainers.pl to fetch from
the MAINTAINERS file the names of folks. That could be imported
in?
> 
> Tim.
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xen.org
> http://lists.xen.org/xen-devel
>

>>> On 12.03.13 at 18:00, Ian Campbell <Ian.Campbell@citrix.com>
wrote:
> With the numbers of patches flying around it can be a bit hard for a
> maintainer to spot patches which (s)he is expected to comment on / deal
> with, especially now that we have a larger number of maintainers of
> varying subsystems etc.
> 
> I think it might be time to start asking patch submitters to CC the
> maintainers of the code they are touching, based on the MAINTAINERS
> file.
> 
> If people (specifically maintainers) agree with this change then we can
> start by updating the "Submitting Patches" docs on the wiki etc
and
> gently start requesting submitters to add the CCs.
I don''t have a strong opinion either way, particularly because the
way my mail rules are set up I won''t pay much more attention to
stuff I''m getting Cc-ed on than to stuff just sent to xen-devel (the
only aspect catching my attention in that case is the double mails
I''m getting, which at times - when xen-devel delivery lags -
isn''t
really meaningful at all). IOW to me this would merely mean a
bigger inbox with not much gain.

And then again, even on Linux I find it quite ugly to have to copy
not just maintainers, but people merely having touched the code.

Plus the general expectation there appears to be to mail patches
_to_ the real maintainer(s), and copy others who touched the code
as well as the corresponding list. That, minus the often pointless
copying of other only possibly interested parties who touched the
code in question, would seem more appropriate to me.

Jan

On Wed, 2013-03-13 at 08:17 +0000, Jan Beulich wrote:
> >>> On 12.03.13 at 18:00, Ian Campbell
<Ian.Campbell@citrix.com> wrote:
> > With the numbers of patches flying around it can be a bit hard for a
> > maintainer to spot patches which (s)he is expected to comment on /
deal
> > with, especially now that we have a larger number of maintainers of
> > varying subsystems etc.
> > 
> > I think it might be time to start asking patch submitters to CC the
> > maintainers of the code they are touching, based on the MAINTAINERS
> > file.
> > 
> > If people (specifically maintainers) agree with this change then we
can
> > start by updating the "Submitting Patches" docs on the wiki
etc and
> > gently start requesting submitters to add the CCs.
> 
> I don''t have a strong opinion either way, particularly because the
> way my mail rules are set up I won''t pay much more attention to
> stuff I''m getting Cc-ed on than to stuff just sent to xen-devel
(the
> only aspect catching my attention in that case is the double mails
> I''m getting, which at times - when xen-devel delivery lags -
isn''t
> really meaningful at all). IOW to me this would merely mean a
> bigger inbox with not much gain.
> 
> And then again, even on Linux I find it quite ugly to have to copy
> not just maintainers, but people merely having touched the code.
> 
> Plus the general expectation there appears to be to mail patches
> _to_ the real maintainer(s), and copy others who touched the code
> as well as the corresponding list. That, minus the often pointless
> copying of other only possibly interested parties who touched the
> code in question, would seem more appropriate to me.
Right, I would only suggest copying the actual maintainers (i.e. those
listed in MAINTAINERS) + anyone who you know *from personal experience*
(i.e. not automated) might be interested in the change. The mad Linux
way of CCing anyone who ever touched the file isn''t useful IME.

Ian.

On Tue, 2013-03-12 at 20:35 +0000, Konrad Rzeszutek Wilk
wrote:
> On Tue, Mar 12, 2013 at 05:07:42PM +0000, Tim Deegan wrote:
> > At 17:00 +0000 on 12 Mar (1363107618), Ian Campbell wrote:
> > > With the numbers of patches flying around it can be a bit hard
for a
> > > maintainer to spot patches which (s)he is expected to comment on
/ deal
> > > with, especially now that we have a larger number of maintainers
of
> > > varying subsystems etc.
> > > 
> > > I think it might be time to start asking patch submitters to CC
the
> > > maintainers of the code they are touching, based on the
MAINTAINERS
> > > file.
> > > 
> > > If people (specifically maintainers) agree with this change then
we can
> > > start by updating the "Submitting Patches" docs on the
wiki etc and
> > > gently start requesting submitters to add the CCs.
> > 
> > Fine by me; I usually work from my xen-devel inbox but I
don''t object to
> > being Cc''d directly.
> > 
> > Can we provide git/hg runes to add Cc:s automatically?
> 
> That would be neat. Linux uses scripts/get_maintainers.pl to fetch from
> the MAINTAINERS file the names of folks. That could be imported in?
Linux''s get_maintainers.pl is way too aggressive about who it defaults
to CCing (basically the world and his dog). How about we go with asking
for manual CCs for now and we can revisit tooling (e.g. tweaking the
get_maintainer''s defaults) once we''ve got a feel for what
works for us.

Ian.

On Tue, 2013-03-12 at 17:00 +0000, Ian Campbell wrote:
> I think it might be time to start asking patch submitters to CC the
> maintainers of the code they are touching, based on the MAINTAINERS
> file.
Based on the discussion which followed I think we now have a rough
consensus that we want to do this.

So we can now *gently* encourage submitters to add CCs based on the
contents of the MAINTAINERS file when replying to patches etc.

I have updated the guide at
http://wiki.xen.org/wiki/Submitting_Xen_Patches
with some the request to do this and information on how to achieve it
with hg email.

I went to do it for git too
http://wiki.xen.org/wiki/Submitting_Xen_Patches_with_Git
but it seems it already recommends CCing the maintainer! I tightened up
the wording to make it more than a suggestion and to keep it inline with
the hg wording.

I''m not sure if there is anywhere else to update. Let me know if you
think of somewhere (or fix it yourself!)

Ian.

Ian Campbell <Ian.Campbell <at> citrix.com>
writes:
> On Tue, 2013-03-12 at 20:35 +0000, Konrad Rzeszutek Wilk wrote:
> > On Tue, Mar 12, 2013 at 05:07:42PM +0000, Tim Deegan wrote:
> > > At 17:00 +0000 on 12 Mar (1363107618), Ian Campbell wrote:
> > > > With the numbers of patches flying around it can be a bit
hard for a
> > > > maintainer to spot patches which (s)he is expected to
comment on / deal
> > > > with, especially now that we have a larger number of
maintainers of
> > > > varying subsystems etc.
> > > > 
> > > > I think it might be time to start asking patch submitters to
CC the
> > > > maintainers of the code they are touching, based on the
MAINTAINERS
> > > > file.
> > > > 
> > > > If people (specifically maintainers) agree with this change
then we can
> > > > start by updating the "Submitting Patches" docs on
the wiki etc and
> > > > gently start requesting submitters to add the CCs.
> > > 
> > > Fine by me; I usually work from my xen-devel inbox but I
don''t object to
> > > being Cc''d directly.
> > > 
> > > Can we provide git/hg runes to add Cc:s automatically?
> > 
> > That would be neat. Linux uses scripts/get_maintainers.pl to fetch
from
> > the MAINTAINERS file the names of folks. That could be imported in?
> 
> Linux''s get_maintainers.pl is way too aggressive about who it
defaults
> to CCing (basically the world and his dog). How about we go with asking
> for manual CCs for now and we can revisit tooling (e.g. tweaking the
> get_maintainer''s defaults) once we''ve got a feel for what
works for us.
linux''s get_maintainers by default no longer cc''s the world.

It looks to see if a file has a maintainer, and if it does, it lists
only that person (and any other higher-level by path maintainer).

If a file does not have a maintainer, it tries to use git or hg
history sign-offs.

These behaviours can be controlled by adding command line options.

--git             (always use git history <default:n>)
--git-fallback    (use git history when there''s no maintainer
<default:y>)

So using get_maintainer --nogit --nogit-fallback might do what you want.