abartlet at samba.org
Tue Jul 5 11:42:05 GMT 2005
On Tue, 2005-07-05 at 20:49 +1000, Andrew Bartlett wrote:
> As Samba4 and Heimdal continue down this little road towards AD-like
> network behaviour, we are starting to depend on having the kerberos
> library around. By the time we ship, Kerberos will clearly not be an
> optional part of Samba4, and at some point we need to figure out how we
> ship the source code.
> Currently, we rely on the developer/alpha tester separately checking on
> Heimdal from lorikeet, and placing it in samba4/source/heimdal. This
> prevents Samba4 from being tested on the build farm with Heimdal, and
> means that some developers test with, and others without the kerberos
> support. This mismatch has caused development disruption.
> What is at issue is how to proceed. The options I see are to:
> - SVN link (reverence, or whatever) the current repository from
> lorikeet into samba4
> - SVN import the entire lorikeet/heimdal repository into samba4
> - SVN import only the files we currently compile into samba4
Now, my preference is for the SVN link, preserving the current history.
However, this has issues with the anoncvs, and with possible conversion
of the repository to future version control systems.
If this option isn't available, I strongly prefer a full import of the
Heimdal tree. Doing so preserves the current situation that allows us
to test our Heimdal tree modifications in a standalone build, as well as
I see this 'test a real heimdal build' as being vital to the ongoing
Heimdal/Samba project. One of the biggest potential costs in Samba4 is
the long-term maintenance of a fork of Heimdal kerberos: As such, I see
it as vital avoid deviating from the 'offical' distribution, and to
ensure that we can push changes upstream.
This has worked quite well over the past, and today I saw both the costs
and benefits of that approach: This morning, I merged lorikeet/heimdal
up to current Heimdal, where lha had merged my libkdc patch. As I
hadn't merged for a while, this was a little painful, but as a result
Samba has a smaller diff to maintain.
Part of what allowed the libkdc patch (and many smaller, earlier
changes) to be merged upstream was the fact that it was built and tested
in a standalone Heimdal tree, as a matter of course. If and when I
committed patches that broke the standalone build, the build farm would
happily tell me so. This approach has reduced the workload for
everybody, and it avoids merging hell...
As we continue to reduce the diff, I actually want to get to a point
where we could do a release taking some official heimdal tarball, and
apply a patch. I feel a partial import may jeopardise this long-term
solution, by making it harder for lha to import our small change into
the upstream tree.
Andrew Bartlett http://samba.org/~abartlet/
Samba Developer, SuSE Labs, Novell Inc. http://suse.de
Authentication Developer, Samba Team http://samba.org
Student Network Administrator, Hawker College http://hawkerc.net
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Size: 189 bytes
Desc: This is a digitally signed message part
Url : http://lists.samba.org/archive/samba-technical/attachments/20050705/50dfde21/attachment.bin
More information about the samba-technical