Should we do a alpha2 now?

[Andrew Bartlett]

On Fri, 2007-12-07 at 09:04 -0500, simo wrote:
> On Fri, 2007-12-07 at 21:46 +1100, Andrew Bartlett wrote:
> > I'm wondering if we should make a Samba4 alpha2 before, or after the
> > python merge?
> 
> After, what use it to release something where part of the stuff is of no
> interest because it is about to be removed?

To allow testing of the rest, without the instability of the new code
impacting on the ability to release, and the testing that users may do
of the rest. 

I want to mark a (reasonably) stable point before the python code
lands. 

> > The code is churning a lot recently, and it's only going to be worse
> > once we merge in the python code.  I'm wondering if we might be better
> > off making a release now, particularly now that MMC Active Directory
> > Users and Computers (ADUC) is better behaved? 
> 
> I see no pressure to make such a release, what would be the reason to ?

The last release didn't work very well with ADUC, and I think it has
issues with group policy.  I'm much happier with many aspects of that
code now.

> > Perhaps we could do the GIT conversion just after that release?
> 
> I'd rather do the git conversion sooner than later, it's becoming a
> stress having to switch between versioning systems.

Given Jelmer's comment, I would like it left until his rather complex
python merge is done. 

> > For those not following the commits, I've been working on things that
> > the ADUC plugin use for group manipulation, such as extended DNs, linked
> > attributes and ranged-results.  The associated fixes also addresses
> > issues seen at the plugfest, where some 3rd party clients applied the
> > 'extended dn' control to bind requests. 
> > 
> > In other changes since the last alpha, we have rolled back SWAT to the
> > earlier version without the AJAX LDB editor, but have added in
> > pre-configuration for phpLDAPAdmin.  
> 
> Didn't we decide to axe swat completely ?

Not quite.  It's still twitching, and I didn't have the heart to make
the final cut.  I'll let Jelmer's python decide if it wants it for
breakfast...

> > With regard to other things I had hoped to see land for an alpha2
> > release, we still don't have CTDB or clustering in Samba4.  That effort
> > appears to have stalled around the release of the Samba3 implementation
> > of that code. 
> > 
> > Any thoughts?
> 
> I think we should wait for python, ctdb and git changes, once those are
> settled, an alpha release might make sense.

That would put it off by another 2 months, I suspect. 

> If you think people need released tarballs to test, then I guess we
> should resurrect the idea of nightly or weekly automatically generated
> snapshots, that we tag a s good from time to time and let live longer on
> the web server.

That could work, but the point I see in a release is in helping people
find working snapshots, not just snapshots.  For the past while, the
tree has often not built for a good portion of the day.  I'm not happy
about that, but I also see particular value in marking some stable
points along the way, if only for external impressions (which do matter,
even if we hoped they would not).

Finally, I'm writing up an article for a magazine, and it's becoming
obvious that pointing users to the unknown future (in particular if the
python code changes the provision process) for a howto section just
won't fly. 

Andrew Bartlett

-- 
Andrew Bartlett
http://samba.org/~abartlet/
Authentication Developer, Samba Team           http://samba.org
Samba Developer, Red Hat Inc.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
Url : http://lists.samba.org/archive/samba-technical/attachments/20071208/1213dd29/attachment.bin