I dont particularly see a great need for such docs as its mainly basic
maven dependency handling long documented elsewhere, and it seems odd
they would be seen as needed now after ~9 years of this. However..
One approach would be to have an explicit or implicit JMS 2 dep and
then actually exclude the 1.1 spec dep from the client, e.g something
that would probably fit the bill might be (untested, just
typed+copied):
<!-- Your JMS 2 related dep here, e.g an API jar directly, or
just another component already bringing it in -->
<dependency>
<groupId>jakarta.jms</groupId>
<artifactId>jakarta.jms-api</artifactId>
<version>2.0.3</version>
</dependency>
<!-- ActiveMQ 5.x client, minus the JMS 1.1 spec dep -->
<dependency>
<groupId>org.apache.activemq</groupId>
<artifactId>activemq-client</artifactId>
<version>5.17.0</version>
<exclusions>
<exclusion>
<groupId>org.apache.geronimo.specs</groupId>
<artifactId>geronimo-jms_1.1_spec</artifactId>
</exclusion>
</exclusions>
</dependency>
Another approach is just having both API deps on the classpath, e.g
brought in by the app deps explicitly, or just another component - if
the stuff being done only needs a 1.1 impl functionality it as likely
wont care about them being there or which is used.
On Mon, 21 Feb 2022 at 13:27, Jean-Baptiste Onofré <j...@nanthrax.net> wrote:
>
> We should at least document how to use ActiveMQ 5.17.0 with JMS 2.0
> from a dependency standpoint (like excluding JMS 1 jar from dep and
> adding JMS 2 as proposed by Chris).
>
> That's the minimum we should do IMHO. We can maybe already update the
> dep in AMQ client.
>
> I propose to wait for Matt's standpoint as he worked on JMS2 stuff (I
> was the initializer ;)).
>
> Regards
> JB
>
> On Mon, Feb 21, 2022 at 2:17 PM Robbie Gemmell <robbie.gemm...@gmail.com>
> wrote:
> >
> > If it doesnt actually implement _any_ of it, then personally I dont
> > think that makes sense, as I've said before. It's of no consequence to
> > the people who have already been using a 2.0 api jar with the 1.1 impl
> > for anything up to 9 years now if that suited them, something which is
> > trivial to do, and it also doesnt really give any advantage to the
> > other users who aren't doing that. It would seem largely just
> > misleading (to me anyway) to add it now all on its own. It has waited
> > this long, it seems better it could wait just a little longer until it
> > is actually implemented in a following 5.18.x series, which neednt be
> > far very away at all once it is actually ready.
> >
> > On Mon, 21 Feb 2022 at 12:58, Jean-Baptiste Onofré <j...@nanthrax.net>
> > wrote:
> > >
> > > Hi guys,
> > >
> > > I agree about the change, and it was not what we agreed on the mailing
> > > list.
> > >
> > > The initial plan for 5.17.0 about JMS 2.0 is "just" to update the
> > > client side to support JMS 2.0 and throw UnsupportedOperationException
> > > for JMS 2.0 specific method.
> > > I think it's good enough for 5.17.0 to support the first JMS 2.0 round.
> > >
> > > More than that (client/broker side) should go in 5.18.0 IMHO.
> > >
> > > Regards
> > > JB
> > >
> > > On Mon, Feb 21, 2022 at 1:35 PM Robbie Gemmell <robbie.gemm...@gmail.com>
> > > wrote:
> > > >
> > > > Finally doing a 5.17.0 release sounds good.
> > > >
> > > > That said, I dont personally think
> > > > https://github.com/apache/activemq/pull/729 is ready for inclusion in
> > > > a release though, even with an '-rc1' adorned version number
> > > > previously suggested but apparently no longer planned, since even as a
> > > > 'first phase' it is surprisingly incomplete, adding some of the JMS 2
> > > > 'simplified API' but not even doing much of the basic JMS 1.1 level
> > > > functionality within it, like setting a MessageListener on a
> > > > JMSConsumer, or creating a durable subscriber (non-shared), or
> > > > JMSContext's acknowledge() method for doing client-ack (presumably the
> > > > message method works though), etc.
> > > >
> > > > It also just seems very odd to even think about just *starting* to
> > > > including stuff like that on main within a couple days of intending to
> > > > do a release thats nearing being *years* in the making, and getting
> > > > describe to users as '2-3 weeks' for way over a year now, including
> > > > multiple times in the last few months.
> > > >
> > > > For me, the most obvious idea at this point would actually be for
> > > > 5.17.x to be branched and proceed without this. Theres a load of stuff
> > > > in it already that is long overdue like the JDK11 build etc. I would
> > > > go so far as to say the prior API jar change from early November
> > > > (https://github.com/apache/activemq/pull/682) should also be
> > > > effectively reverted, it makes no sense to me on its own. Then all of
> > > > this stuff then worked on towards a 5.18.x release that actually
> > > > implements and tests things to a reasonable level thats less likely to
> > > > see even trivial use cases fail to work.
> > > >
> > > > Robbie
> > > >
> > > > On Mon, 21 Feb 2022 at 04:55, Jean-Baptiste Onofré <j...@nanthrax.net>
> > > > wrote:
> > > > >
> > > > > Hi guys,
> > > > >
> > > > > I worked on the log4j2 update PR this weekend, fixing almost all unit
> > > > > tests using a custom appender. I just have to fix the
> > > > > activemq-web-demo test and squash, and the PR will be good to be
> > > > > merged. I will do that today.
> > > > >
> > > > > Then, later today and tomorrow I will work on using jetty modules
> > > > > instead of jetty-all and update to Jetty 9.4.45.
> > > > >
> > > > > I will do a pass on Jira and PRs, especially the ones from Matt. @Matt
> > > > > can you please ping me on slack to check together the status of the
> > > > > PRs ?
> > > > >
> > > > > Regarding this, I would like to submit 5.17.0 to vote this Thursday if
> > > > > there are no objections.
> > > > >
> > > > > Regards
> > > > > JB
