Re: [RFCv2 00/15] RFCv2: Consolidated userspace RDMA library repo

[Leon Romanovsky <leonro-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>]

[-- Attachment #1: Type: text/plain, Size: 3322 bytes --]

On Tue, Aug 30, 2016 at 10:30:33AM -0600, Jason Gunthorpe wrote:
> On Tue, Aug 30, 2016 at 10:35:21AM +0300, Leon Romanovsky wrote:
> > On Tue, Aug 30, 2016 at 12:17:16AM -0700, Christoph Hellwig wrote:
> > > On Tue, Aug 30, 2016 at 09:08:42AM +0300, Leon Romanovsky wrote:
> > > > Doug is a busy person and there is a limit on how fast he can handle it.
> > > > He is already buried under his internal and external responsibilities.
> > > >
> > > > Placing him responsible for vendors code will add extra step,
> > > > extra complexity to the chain and will hurt kernel/libibverbs flows.
> > >
> > > Leon, calm down.  What Jason is proposing is to apply the exact same
> > > flow we have in the kernel to a much smaller project.  We prove it
> > > works on a giant project, and it will work on a smaller one.
> >
> > Kernel project has more than "pile of code in one place".
>
> Sigh, that is insulting Leon. I haven't just pulled a bunch of tar
> files together. Go look at the commits, and actually try out the
> repo. I did a lot of work to harmonize things and there is still more
> to do. I just pushed another update for you which includes some of the
> cruft removal.

"Insulting", you are taking it to far. No one is arguing with you about
the complexity of what you had done. I truly appreciate the effort which
you invested in it, but without the process it is not efficient as it
supposed to be.

>
> Besides, how do you think the kernel got to where it is today? Do you
> think it started out like this? NO. This is a process, starting with a
> consolidated repository, getting the distros on board, getting a team
> willing to even look at this stuff together, developing a process and
> so on.

So let's define a process together.

>
> The first step is the 'make it easy step'. Today everything is just
> too hard for developers. When Christoph Hellwig says building our user
> space is too hard *you should listen*. He isn't wrong, and he isn't
> inexperienced at this.

Sure, I listen, I was in that situation a year ago and still
remember how much time takes to setup the development machine,
but you should read second half of Christoph's answer to me too -
about overwhelmed maintainer in RDMA stack. It is enough to release
that valve "to push the train" and improve the user space domain.

>
> How many potential community members have been dissuaded by that
> simple fact??
>
> > > Maintainer overload always is a problem, and it's helped by adding
> > > more maintainers.  And yes, both the kernel RDMA stack and the user
> > > code should have a small maintainer team, but that's a different
> > > discussion.
> >
> > Awesome, let's start from this discussion to clear the ground and move
> > to technical proposals after it.
>
> We've already been over this ground, multiple times over
> years. Consolidation is the techincal direction a lot of people want
> to move toward.

Again, all participants are agree, the disagreement is in
preparation step before such move.

>
> Whatever maintainership structure is decided on can steward the
> consolidated repo.
>
> Jason
> --
> To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
> the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 819 bytes --]