[Debburn-devel] What is going on?

[Peter Samuelson]

[Albert Cahalan]
> I still haven't been added to the project (albert-guest),
> much less given the ability to commit changes.

The two are the same thing - there's no separate tier of "project
member who doesn't have svn access".  The current policy is not to give
out commit access until a person has contributed a few patches so his
ability and judgment can be measured.  Joerg and Eduard already knew
some of us, but obviously they aren't likely to know everyone who
wanders in to the project.  The alioth security model does not provide
a way to give commit access without giving direct filesystem access to
the repository (i.e., the ability to vandalize or trojan it
arbitrarily), so a bit of caution is necessary in handing out access.

If you submit a few good patches, I am sure you will soon be added.  If
necessary (and I think it won't be necessary), I'll vouch for you
myself on the basis of having seen you in Linux kernel development
circles since the mid-to-late 1990s.  Even if you did advocate devfs at
the time.  (As did I.)

I think it's that way in most open source projects, really.

> Do we have any non-Linux developers or anyone willing to give out
> non-Linux accounts? Do we care?

Now that I don't know.  It would also be nice to nail down specific
portability goals.  We've already decided that ANSI C is a reasonable
requirement, but can we assume all cdrkit users will have 4.3BSD or
SVR4 functionality?  It would simplify life considerably if we could
require a C99 compiler and a POSIX runtime, but I'm guessing that is
asking too much.

> Stable releases need to go on branches so that normal development and
> version history won't get ugly.  Excepting well-defined projects
> involving long-term multi-developer hacking that is likely to
> destabilize things, it's best to avoid doing the mainline development
> off of the mainline.

I agree with this approach, in general.  I think the reason we're being
conservative right now is that we'd like to get a release out with full
parity with Joerg Schilling cdrtools (I think we're nearly there), call
it 1.0, _then_ move on to potentially destabilizing changes.  Sure, we
could create a stable branch now, but I think with the current focus on
producing the first stable release, it would result in a lot of merging
of small stuff.

> BTW, we could use a more informative web site.
> This would be nice:
> 
> a. user info: tarball download, online documentation
> b. dev info: SVN info, mailing list
> c. tech docs: links to various standards and OS info

Joerg now has control of cdrkit.org, though he apparently hasn't had
time to actually configure it as a separate virtual host yet.  In the
long run, I'd like to see the website itself maintained in svn, as is
done with many other projects (including Subversion itself - compare
http://subversion.tigris.org with the copy maintained in their
repository, http://svn.collab.net/repos/svn/trunk/www/).

Peter
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : http://lists.alioth.debian.org/pipermail/debburn-devel/attachments/20060910/aacbba9b/attachment.pgp