Hi David,
I actually fully agree with your statement in principle.
Personally, I
would be all for it, we did the same kind of rewrite in Camel when we
moved from 1.x to 2.0, and it was a long and painful process.
Speaking
of which there were talks about a Camel 3.0 for at least 3 years that
I am quite skeptical of.
In this case, hornetq is actually a completely different product,
voted into the community by a vendor who has the vast majority of the
pmc votes. Even that would not matter that much if the rest of the
community would buy into the vision of hornetq morphing into the next
amq, more or less a drop in replacement as others stated in the
thread.
Your analogy with Apollo is exactly what I mean. As much as I like
the
elegance of scala I did not buy into its ability to catalyse a
community. If enough users jump off the activemq 5.x wagon and its
successor doesn't get enough steam, then activemq would reach a "dead
end".
I don't know what the future would bring. And honestly, I don't know
what the best choice is. Based on my previous experience, I choose to
err on the conservative side, yet I wish for the best.
The onus is on the new podling to build a community. The fact that it
was adopted by the activemq project doesn't mean that it must use the
same name, it means that the activemq pmc took on the duty of
mentoring the new committers (past example, among others: smx kernel
moving to felix and then going tlp).
Speaking of mentoring, it is not something that can be imposed.
Personally, I have never been asked anything by new committers in the
activemq community. I do however mentor other communities. Tinkerpop
is such an example, and man, it's a joy, the project is growing well
and the technology is awesome. There are other examples where it
doesn't work so well.
Bottom line, activemq6 must build a community. With a different name
it could prove that it's capable of doing it on its own. This way
it's
stealing from the activemq5 community. That's fine, but then address
its needs and make the users happy: seamless migration, near drop in
replacement.
Cheers,
Hadrian
On 03/23/2015 10:43 PM, David Jencks wrote:
It seems to me that a different name would mean a different project.
IIUC AMQ accepted the code so not having it turn into amq-xxx
seems a
bit odd to me.
What is stopping all the non-jboss-employees (yes, showing my age
here) from enthusiastically digging into the code and adapting from
amq 5 or implementing anew all the missing bits? My limited
understanding is that the amq 5 broker sort of reached a dead end
and
needed a rewrite rather than incremental improvement, first Apollo
tried to do it in Scala which not enough people understood, and now
there's a new bunch of java code to look at.
It seems to me that one of the roles of the preexisting
committers is
to help the new ones learn about apache. What better way than by
pitching in and working on the code together?
wishing I had time to actually contribute rather than just argue….
david jencks
On Mar 23, 2015, at 9:28 PM, Hadrian Zbarcea <[email protected]>
wrote:
Now here lies the problem.
I agree that it captures the intent well. That also creates an
expectation from the users and sort of a promise from the activemq
pmc, amplified by the vendors' marketing (well, exactly one in this
case). The same promise has been made with apollo.
I am less concerned with the rewrite. To me that is not an
issue. If
smaller or larger parts are rewritten but maintain (reasonable)
feature parity, it is an evolution of the same project.
I am however more concerned with the ability of the activemq6
podling/subproject to build a diverse community. So far I don't see
encouraging signs. My fear is that the result will be alienation of
the more diverse activemq 5.x community (still less diverse than it
should be) and turn activemq into a one company show.
So far it looks it looks to me that the perception card was played,
with the choice of name. It *sounds* like activemq6 the
evolution of
activemq. How will the current pmc ensure that this is really gonna
be the case? (fwiw, I do get questions about the relationship
between amq6 and 5 already, and for the life of me I don't know how
to answer).
Choosing a different name, as I think Rob suggested too, would have
made this a moot point.
My $0.02,
Hadrian
On 03/23/2015 10:07 AM, Gary Tully wrote:
+1 to the -M1 naming, I think that captures intent perfectly.
On 23 March 2015 at 10:09, Andy Taylor <[email protected]>
wrote:
So I think the consensus is to go with ActiveMQ 6.0.0-M1 so we
will go
ahead and cut a new RC in the next day or so. We will also add
some
content the website so users are clear that currently there isn't
feature parity between ActiveMQ 5 and ActiveMQ 6. We will then
raise
JIRA to map out a migration path post release.
On 20/03/15 20:40, Clebert Suconic wrote:
On Fri, Mar 20, 2015 at 4:25 PM, artnaseef <[email protected]>
wrote:
Please help me to understand how this would go.
We would use 6.0.0-M1, 6.0.0-M2, etc until when? Until we are
ready to
declare that 6.0.0 is a replacement for 5.x?
After that, then we simply drop the -M# (i.e. release the first
6.0.0)?
Yeah.. That's exactly how I see it.
