psychology of version numbers

[Dave Morrison]

Hi all,

Is there any consensus on what work, functionality and testing is needed before
arla can be bumped up to a 1.0 release?  Is the idea to have both client and
server side rock-solid and usable for daily work in production systems?  Would
it be OK, instead, if only the client side seemed solid?  Or if it still had
known problems on some systems?  There are a lot of points listed in the TODO
file, but I don't have any sense of whether all the items in there are things
that folks feel are necessary to have finished before replacing the "0." with
"1." in the arla version number.

On the systems I work with (linux, kernel 2.2.14), the last few arla releases
have been quite usable.  _Of course_ I realize that this only represents a tiny
fraction of all the systems on which people use arla.  For all I know, there
might be such severe problems on other platforms that it's reasonable to stick
with the "0.x" numbering for now.

The reason I ask all this is that when I suggest people use arla, they see a
version number like 0.30 and they get a bit pale, a little queasy, slightly
wobbly, and wonder if I'm pulling a fast one on them.  I've used arla since
about version 0.9, so I know true danger when I see it, and I don't think the
current code is all that threatening.

Cheers,
Dave
-- 
David Morrison  Brookhaven National Laboratory  phone: 631-344-5840
                Physics Department, Bldg 510 C    fax: 631-344-3253
		          Upton, NY 11973-5000  email: dave at bnl.gov