ant-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jason Hunter <jhun...@collab.net>
Subject Analyzing Bessie (Was: Whoa Bessie...)
Date Wed, 20 Dec 2000 03:44:32 GMT
I've been captivated by this list traffic the last few days.  It's
fascinating data, and useful for my day job where I advise companies
about how open source works.  This issue is pushing open source ideas
into areas where no one has a lot of experience and where there aren't
any well defined rules.

Now's my time to jump in.  As someone who's followed Ant since the early
days, here's what I see on the past, present, and future of this issue
(yes, I did read "A Christmas Carol" over the weekend).

James originally created the Ant tool within Sun as a way to build
Tomcat.  He needed a cross platform build mechanism for the server he
was creating, and he recognized that Make sucks.  All of us knew this,
but unlike most of us who created hacky Makefiles, he invented something
innovative: Ant.  He named it Ant because it was a little thing that
built big things.  :-)  Ant was his baby more than Tomcat really, and
while he fought to open Tomcat he also had an eye to open Ant along with
it.

Ant in the open was justifiably recognized as a cool build tool, and we
all scrambled to use it and some of us to improve upon it.  I personally
have dropped Make for good!  :-)  Around this time James got drawn away
on other things (finishing the Servlet API spec, cleaning up the XML
JAXP spec) and couldn't help much with Ant.  So the community here built
up around the tool and *rightly* feel like it's partly theirs now.

(You can see the conflict coming, can't you...)

James takes a new job within Sun that allows him to dedicate some real
time to Ant.  He wants to jump back into his old position as "vision
thought leader".  However, the community that has grown around Ant
doesn't recognize him as such, because in the time since they joined he
hasn't been very active.  So they want him to prove his worth again with
the advice "start contributing, help with patches, start advising more
on ant-user and try to be more constructive with your criticism."  He
feels like George Washington returned from afar welcomed to the nation
he helped found with not a "wecome back" but rather a "prove your
patriotism again".  And of course, that hurts.

(Yep, there's the conflict...)

The good news is that progress is being again on debating the technical
merits of each proposal.  That's what we're all best at.  :-)

But there's one open issue.  What do we do with the name "Ant"?

The "Rules for Revolutionaries" (created by James ironically) doesn't
address this issue.  The thought was that there would be consensus on a
path to take and that the name would follow that path.  Here's a
situation where we have what may be fundamental disagreements about
design, and there may have to be *named releases* before we get any
consensus, if we ever do.  What I think has concerned James is that he
wants some guarantee that if he believes his approach is technically
superior to the competition, he's OK with the competition but doesn't
want to lose the name "Ant" to what he would consider a "fork" from his
original vision.

In Apache, afterall, a name is very important.  The Apache name can only
be used by derivative works with the express permission of the ASF.  If
someone else thinks they can do better, the code is theirs *but not the
name*.

This situation with internal forks is hard because who gets to have the
fork with the original name?  I think in the situation where such a
decision needs to be made it should be decided by the overseeing PMC. 
Name choosing is one of the standard duties of a PMC (although usually
only selecting initial names!).  Furthermore, I would hope that (as with
child custody) they try to leave the name with the original creator
wherever possible.  If not, it's a strong disincentive to host a project
with Apache:  You could create a project and have its "brand" taken by a
fork you disagree with.

Apache is about collaboration, and we share our code internally as well
as externally.  But we preserve the name as our brand.  If someone
outside Apache forked Ant, they'd have to choose another name.  The
question now is do we offer the same protection within Apache?

-jh-

Mime
View raw message