[OpenAFS-devel] Re: Making supergroups default to on

[Russ Allbery]

Brandon Allbery <allbery.b@gmail.com> writes:

> My personal worry would be related to the other recent thread asking
> about issues when using non-supergroups pts on a supergroups pts
> database; I would expect that there be some way to at least detect the
> incompatibility, and if not do something saneish then at least avoid
> causing corruption.

In the long run, I don't think supergroup support should be conditional.
(I have no opinion over whether preventing people from creating them
should be conditional, but if it is, it should be a run-time condition.)
So one way to fix this problem that might be easier than fixing the
current conditional code would be to properly merge the supergroups code
and eliminate the preprocessor defines so that we don't have two parallel
versions of the code.

Then, adding supergroups is a one-time upgrade issue that will change the
database in a way that isn't backward-compatible, which while not very
elegant could just be documented as such.

The reason why having non-supergroup code recognize and not corrupt
supergroups is important right now is that it's possible to switch back
and forth (it's just a bad idea).  It's reasonable to not support
downgrades, but not reasonable to not support turning an option off and
on.

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