Paul Blackburn <[EMAIL PROTECTED]> writes:
> Is there a process for defining new @sys?
I think there should be. I'd like to see a policy with the following
characteristics:
* Public. A public listing of all machine to @sys mappings, regardless
of whether there's an available AFS client for that system, put on a
web page somewhere.
* Reasonably quick. It shouldn't take a long time to get a mapping for
a new architecture.
* Apolitical. Basically, my ideal world would be for someone to just
pick names, based on whatever heuristics that they want to use, and
not accept a bunch of arguing about it. I really don't care *what*
the @sys name chosen is; I just want there to be one without having to
do lots of bickering for various alternatives.
If IBM would pick up the same set of listings, that of course would be
ideal, but if necessary it's possible to work around that with fs sysname.
I *would* like to see @sys names chosen on more criteria than just what
version of AFS client is associated with that system; for example, I'd
like to see rs_aix42 and rs_aix43 split. One of the most useful things
that @sys can be used for if done right is central software trees. Unless
there's some major reason why we *can't* support that application, I'd
really like to see an @sys mapping database that does support it.
(The fineness of the grain is open for debate, of course; I'm personally
happy to not distinguish between 32-bit Solaris and 64-bit Solaris and use
other mechanisms to handle those programs that care, but that's just my
personal opinion.)
--
Russ Allbery ([EMAIL PROTECTED]) <http://www.eyrie.org/~eagle/>
