directory-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Trustin Lee" <>
Subject Re: DNS Protocol (Was: DHCP Protocol Home)
Date Wed, 20 Jun 2007 01:44:54 GMT
Hi Brian,

On 6/20/07, Brian Wellington <> wrote:
> On Tue, 19 Jun 2007, Trustin Lee wrote:
> > On 6/19/07, Stefano Bagnara <> wrote:
> >> >> What if there were a "frequently updated" dnsjava, containing all of
> >> >> useful community-contributed patches (including dnsjnio), released
> >> >> under a
> >> >> BSD license - would that be a good solution?
> >> >
> >> > Absolutely.  That is the best of the best IMHO.  WDYT guys?
> >>
> >> Sorry but I don't understand the proposed scenario.
> >>
> >> I think the problem is that we need a community around a new dns project
> >> because we want to provide fast respose and more evolution than dnsjava
> >> currently offer.
> >
> > It is true if Brian doesn't have any will to work closely with us.
> > He's not replying to us so far, so I agree with you.  And I don't
> > think what Alex is suggesting to us is not so far from what we have
> > been discussing.  He didn't say it should be outside of the ASF. ;)
> To be honest, I'm not really sure what the intent here is.  For the past 5
> years or so, there really hasn't been much need for much evolution or fast
> response from dnsjava - there haven't been many features that I've wanted
> to add, and there haven't been too many external patches.
> The dnsjnio patch set is, as far as I know, the only code that should be
> part of dnsjava but isn't.  That's mostly my fault - I haven't had much
> time to work on dnsjava recently, and evaluating and working with a large
> patch takes more time than I've had.

The intent here is about build a community around DNS protocol
implementation.  I agree with you that dnsjava is a very mature and
stable product.  But there are also demand for asynchronous DNS
implementation.  dnsjnio is a good example, and some people want to
switch dnsjava's internal implementation with existing NIO framework
to fully utilize asynchronous API.  These changes will need quite a
lot of work and we thought dnsjava and dnsjnio are a great starting
point.  I'm not sure you have interest in these changes, but there are
about four guys who are interested in these changes, so we might be
able to create a nice community around it.  We also thought starting
this work in the ASF will be beneficial to many parties such as James,
MINA, and Directory project because they want to work closely with the
DNS protocol implementor in the near future (or right now).  Is your
question about our intent clarified?  WDYT?

> >> What about the "frequently updated dnsjava" you talk about? Who will do
> >> that? In what environment? What community? As it has been proposed to be
> >> BSD then it cannot be hosted by ASF because the Apache Foundation will
> >> only create/maintain ASLv2 projects (and will also need copyrights for
> >> everything maintained).
> >>
> >> Indeed, I really would like to know whether Brian has objections to such
> >> an effort, what would be his own preferences about dependencies (or
> >> evolutions) of his code, if he would be interested in collaborating in a
> >> more "community oriented" project, if he would be interested in an
> >> oversight role.
> If someone wants to fork dnsjava, I can't stop them.  A "community
> oriented" project sounds like a good idea, but as I said above, there
> really hasn't been much call for new features.  In the 8 years or so since
> the first dnsjava release, there's only been 1 serious request for commit
> access to the repository, which makes me wonder where the community is.
> People occasionally pop up and ask for large features (some of which make
> sense), but very infrequently contribute any code.
> If the decision was made to fork or evolve the code, I wouldn't object,
> but I don't know how much of a role I'd be willing to play.  I'm not using
> dnsjava in any other large projects, so the only real development I do
> (other than applying patches) is adding small features needed for
> miscellaneous DNS testing and adding new record types.  The main reason I
> use dnsjava for testing is that I know the API pretty well, and it's easy
> for me to write code using it - if the API evolved to the point that
> wasn't as familiar with it and my code didn't work, I'd probably continue
> to use the old version or use something else (like dnspython - dnsjava and
> small programs using it are the only java code I write anymore).

Our primary goal is not about forking dnsjava.  I think it's our last
resort.  Especially, I appreciate your effort to maintain dnsjava
project as a previous user and a fan.  With a bigger community, we
could cope better with such a big request because it's not only you
but all committers will have more chance to consider about the worth
of the request.

> > dnsjava is licensed under BSD-license, so we can fork it without prior
> > permission, though we need to try.  (But he's not responding.)  And
> > starting from the ADS DNS protocol provider is also a good idea.  I am
> > not a DNS expert so it's up to the actual contributor of the project.
> > If Alex is interested, it's a great news, and starting the project
> > from the incubator might be a better idea to attract more interested
> > non-committers such as Alex.  Even if we have little code base so far,
> > we could discuss about this project with the incubator PMC and ask if
> > we can start from existing code that we didn't write (dnsjava) due to
> > its inactivity.
> I can't really comment on this, as I don't know much about the ADS DNS
> protocol provider.  Again, though, the "inactivity" you're referring to is
> mostly the lack of any necessary changes.  dnsjava is pretty standards
> compliant.  Other than a good asynchronous interface, there really isn't
> anything that's missing.
> Of course, the inactivity could be the lack of frequent releases to pick
> up minor bug fixes.  I don't do releases more often because releasing
> software on sourceforge is a pain (doing anything with the file release
> system takes many times longer than it should), but the code in CVS (which
> would have been moved to SVN a year ago if sourceforge's SVN conversion
> scripts worked) should be pretty stable, if anyone would like to volunteer
> to do more releases.

I admit 'inactivity' here is incorrect and apologize from my heart if
it dismays you.  Most of all, please consider our intent as serious
will to improve the best-of-breed DNS implementation.

what we call human nature is actually human habit
PGP Key ID: 0x0255ECA6

View raw message