[OpenAFS] Funding the formation of an OpenAFS Foundation

Russ Allbery rra@stanford.edu
Fri, 28 Sep 2012 11:27:42 -0700

It's probably worth being clear, just in case anyone missed it, that I
stepped down from being an Elder and Gatekeeper a while back, and all of
my comments on this thread are purely as an individual.  I'm intentionally
dropping some of the filters that I was applying when I was formally part
of project governance and being a bit more direct than I probably
otherwise would be, since I'm less worried about having everything I say
taken as a statement on behalf of the project.

So y'all should also update the filters with which you read me.  :)

Ted Creedon <tcreedon@easystreet.net> writes:

> If IBM wants backward compatibility they should pay for it. They're
> coasting on your work.

> What's in a name? Who cares?

I was one of the people who from time to time suggested that we just
rename AFS to get away from the trademark issue in discussions among the
Elders.  So to some extent I'm sympathetic.

There are some strong arguments against doing so, though.  AFS has quite a
lot of name recognition outside the set of people on this mailing list.
There are decades of mentions, discussions, blog posts, and so forth out
there in the world, articles in Wikipedia, mentions in file system
comparisons, and past experience with AFS by everyone who has graduated
from Stanford, MIT, CMU, UMich, etc.  There's a lot of name recognition
that we lose by changing the name, and given that one of the things we're
struggling with is convincing people that AFS is a viable enterprise file
system, losing that name recognition is a problem.

But, somewhat more significantly, it's not like we're in a position to
increase the pace of development all that much by abandoning the things
that IBM wants but that other people don't care about.  Sure, we could
take kaserver out of the tree (which would actually hurt a few sites other
than IBM), but that isn't really going to help people develop new code
faster.  It simplifies things a bit, and removes some old code, but it's
not that huge of a change.  Most of the things that IBM cares about we've
been told are important by many other people in the community as well,
such rxkad support remaining until people have a chance to migrate to
something else.

I think if we were in a position where there was something that we could
actually ship, or at least feasibly develop in a short time frame, that
IBM were blocking, the situation would look much different.  But it's far
from clear to me that this is the case.  Most of the problems that people
want to solve by breaking backward compatibility probably shouldn't be
solved that way for reasons entirely separate from IBM's concerns.

Russ Allbery (rra@stanford.edu)             <http://www.eyrie.org/~eagle/>