[OpenAFS-devel] Re: Moving Forwards

[Andrew Deason]

On Mon, 10 Sep 2012 08:56:15 -0600
Ken Dreyer <ktdreyer@ktdreyer.com> wrote:

> 1) Does anyone know of any blocking issues that would prevent us from
> cutting a release from HEAD on 1_6_x right now?

I don't think you ever want to cut from HEAD directly, and certainly not
right now (libtool migrations and whatnot). 1.6.x is possibly waiting
for the linux-symlink-mntpt thing (#130273) to settle a bit, but we may
be 'there' now. Otherwise, I'm not aware of critical issues, and there
are a few important fixes on that branch that some people would like to
see in a release.

> 3) What is the policy (official, or conventional) for getting
> backports into 1_6_x? I have cherry-picked several of Marc's commits
> for newer kernels from master to 1_6_1 in order to get 1.6.1 to build
> on Fedora 18/19 for RPM Fusion. I imagine that these patches haven't
> been backported to 1_6_x at this time because Marc / Gatekeepers don't
> have the time. Is it ok if I just submit my cherry-picking efforts to
> Gerrit against 1_6_x myself, or is that going to interfere with some
> process that the Gatekeepers already do?  I would love to see these
> get into 1.6.2 or 1.6.3.

A cherry-pick is pretty much the same as a code submission; anyone can
and is welcome to do it. At least, I assume that is the case, because
nobody complains when I do it :) Typically the people that will do it
are:

 - The person that made the original submission ("I also want this in
   1.6/1.4/etc")
 - Someone else that hit the same issue fixed by it, and wants it in a
   stable release
 - Derrick, when he pulls like 100 at once (okay, probably not _quite_
   so many at once anymore... I hope)

> Additionally I have a question about infrastructure. OpenAFS hosts a
> *lot* of infrastructure compared to many open-source projects. Given
> the discussions about available resources, should the new ticketing
> system be hosted by OpenAFS, or would GitHub's issue tracker or Google
> Code's issue tracker be viable?

Not having bidirectional e-mail interaction is going to sound like a
blocker for many people. Or at least, it's going to make it really
really annoying to use, to the degree that it may get used even less
than the current system by developers. (Some of those may have proper
email support... I just mention that because that's usually what's
missing from such services, iirc.)

-- 
Andrew Deason
adeason@sinenomine.net