Mike,

While I agree with most of what you state, I fail to see the relevance.

When a user upgrades from project FOO version N to FOO version N+1, there is an expectation of reasonable backwards compatibility. Version N+1 may or may not be a complete rewrite, but rules of engagement are clear. The intent for Artemis to become ActiveMQ 6 was already stated ages ago.

Since RH was mentioned there is ok to completely replace the guts of a project by keeping its name and bump the version number, with whatever that means for users (more PS for upgrades, etc). At the ASF we don't do that. This went all the way to the board and the result was clear. Looks like it may end up there again.

So, -1.
Hadrian

[1] https://semver.org/

On 12/06/2017 02:08 AM, Michael André Pearce wrote:
Based on the Dev discussion linked I believe this vote was more making the 
direction and future clearer for users, its not deprecating overnight 5.x, but 
simply clearing up what is ActiveMQ 6 going to be.


On your commends about JBoss.

I don’t think vendor versions should come in here. Apache projects and its 
versions should have their own lifecycle not influenced by what vendors 
re-packing and supporting apache projects are doing. This is an Apache Project, 
NOT a RedHat/JBoss project.

Many other apache products which have vendors releasing their own versions, 
such as:

Apache Hadoop (HDFS) with Hortonwork, Cloudera, MAPR
Apache Kafka with Confluent
Apache Ignite with GridGain

They all have versions that conflict and/or are different with the upstream 
Apache projects.

On that note re your comment ""JBoss AMQ 6" is Apollo" whilst I’m not a RedHat 
person/employee so I cannot be an official source (I work for a company that uses both ActiveMQ as 
some of its message brokers), but from their documentation available publicly on their site, JBOSS 
AMQ 6 is based on ActiveMQ 5.X.

Saying this and re-iterating my previous comment, Apache versioning should be 
agnostic to what vendors are versioning and shouldn’t come into this discussion 
IMO.

On that note to the same cord, i think it may answer a little your question re 
adoption if RH are releasing their vendor product based on it switching from it 
seems 5.X to Artemis shows that the maturity/adoptions of Artemis, they would 
obviously have customers using it, and others transitioning from their previous 
version.

Whilst on Adoption, I’m aware that:

* Spring Framework already has support for ActiveMQ Artemis, its one of the 
options within Spring Boot, along side Rabbit, Kafka and ActiveMQ 5.X 
(https://docs.spring.io/spring-boot/docs/current/reference/html/boot-features-messaging.html)
* WildFly is using it reading their docs 
(https://docs.jboss.org/author/display/WFLY10/Messaging+configuration)
* Other open source projects are building / adopting on it:
  * OpenIoE -> https://github.com/scorelab/OpenIoE
  * Enmasse.io -> http://enmasse.io

Cheers
Mike










On 6 Dec 2017, at 03:51, artnaseef <a...@amlinv.com> wrote:

-1  I think we need to slow down.

While the referenced discussion opened the possibility of unifying on a
single broker, there's a lot more to discuss before that decision is made.
Naming Artemis as ActiveMQ 6 implies to the community that we are
deprecating AMQ 5 now.

For example, the assertion that "I think all the features are covered at
this point" shows a lack of clarity itself.  If we were truly methodical,
then we would have a list of criteria needed for Artemis to take the name
ActiveMQ 6.

ActiveMQ is an important asset to the communities it serves, and it deserves
the greatest of attention and care.

Questions coming to mind for making this decision:
* What is the full list of features needed?
* How much adoption does Artemis have?
* How stable is Artemis?
* What features will be dropped?  Scheduler?  HTTP endpoints?  ...

Just today I ran into the following bug the hard way:
https://issues.apache.org/jira/browse/ARTEMIS-1022.

Notice it's still open after more than 8 months.  It impacts OpenWire
support, which is critical to me as we want the most straight-forward
transition for customers as possible.

Please start to enumerate these points.

BTW, on the confusion front, since "JBoss AMQ 6" is Apollo and "JBoss AMQ 7"
is Artemis, I think renaming Artemis to ActiveMQ 6 will create even more
confusion.

ALSO - one big point.  This DEV list is hard to follow now thanks to the
vast majority of messages being commit messages, and while I 100% agree with
having this discussion on the DEV list, the PMC needs to be made aware of
these discussions and votes on the PMC list.

I'll post the link there now.





--
Sent from: http://activemq.2283324.n4.nabble.com/ActiveMQ-Dev-f2368404.html

Reply via email to