[OpenAFS-devel] Road map, was Proposal for capabilities support in Unix client 1.4.x

[Jeffrey Altman]

Matt W. Benjamin wrote:
> I thought that the goal was to remove the known blockers.  Then a 1.6 could ship with DAFS available but not configured by default, the assumption being that viced was acceptably stable in that configuration.  It would be nice to start talking about how to better validate with DAFS enabled, certainly.
> 
> Matt

The question is "what is the goal of 1.6?"   There are not many
functional feature changes on the HEAD minus DAFS and DAFS is
something that many sites want.  They have been listening to
talks about how great it is for the last 2.5 years.

>From my perspective if 1.6 is going to ship it must be capable
of being run with DAFS turned on even if it is off by default.
Both 1.6-DAFS and 1.6+DAFS must be tested and determined to
be stable.  I do not believe that removing the DAFS code at
this point is a viable option for 1.6.  Nor do I believe that
leaving the code in the tree and disabling it will buy us very
much considering the large impact the addition of the DAFS
support has had on the source tree.

It is unclear how stable 1.5 is with or without DAFS.  We
know that 1.4+DAFS is being used at Morgan Stanley based upon
the statements at the 2008 Workshop.  We also know that
attempts by several individuals and institutions over the
last year to deploy 1.5 with or with DAFS have gone badly.
Many patches have been applied to the tree since then.
Convincing sites that were previously burned to start testing
again is always a challenge.

Jeffrey Altman