samba development

[Cole, Timothy D.]

> -----Original Message-----
> From:	Charles N. Owens [SMTP:owensc at enc.edu]
> Sent:	Wednesday, August 16, 2000 10:09
> To:	Multiple recipients of list SAMBA-NTDOM
> Subject:	Re: samba development
> 
> I agree with this sort of approach whole heartedly.  For a good example of
> how
> this dual branch development approach can work very well take a look at
> the
> FreeBSD Project.  Their use of "Stable" and "Current" branches would, I
> think,
> work very well for Samba.  How ever it is done, the main thing is that the
> development approach be formalized and (gasp) documented, including
> policies
> about when and how bits should be moved between branches.
> 
	Well, this was kind of the idea with TNG versus HEAD, minus some of
the formal policies about moving code (but formal policies don't resolve
design disputes anyway).  The original plan was something like HEAD would
become Samba 2.2, and then TNG would be merged back to HEAD for Samba 3.0.
This slowly changed.

	The major problem (as I see it) is that there are a lot of major
differences on design issues between the "stable" and "current" maintainers
(luke/elrond acting in the latter capacity), and so that not that much code
is moving back to HEAD.

	I believe someone on the a week or two ago implied that TNG was now
a "reference implementation" which would be used more or less only as a
reference point to implement similar functionality in HEAD.  Whenever
exactly that happened, it effectively meant that Samba had forked.

	In the face of this I don't think Luke saw a true fork as being
particularly productive, but he ran out of energy trying to find other
resolutions.

	Maybe I'm talking out of my elbow, though.  I can't pretend to have
been particularly closedly involved with the nitty-gritty of TNG or even
Samba development in general (modulo some abortive attempts at ACL support
and a libsurs).  I don't really have anything like Luke or Andrew or
Jeremy's perspective on this.