ARTEMIS-33 Generic integration with SASL Frameworks

[laeubi]

This adds the opportunity to register new SASL schemes via the default
java service-loader mechanism.
Implementors have to provide an implementation of the ServerSASLFactory
that is responsible for providing instances of the actual scheme.

[laeubi]

I have prepared an example that builds up this new feature to support SASL-SCRAM.
Would be really helpful if this could go into the release next week.

[gtully]

this is great.
could you add a little test for this such that it is exercised without any external dependencies, that will protect it into the future.
Not sure how much work is in that, but maybe even verifying the mechanism is loaded via something like: org.apache.activemq.artemis.tests.integration.amqp.AmqpInboundConnectionTest

[gtully]

It can be any dummy SASL thing, just that it completes the round trip from being requested by a client to being found and loaded by the server.

[laeubi]

@gtully what do you think about transforming the both defaults (PLAIN, ANONYMOUS) to the new way of providing a SASL-Mechanism? Would this suffice?

[laeubi]

I have updated the PR but can't see why the build fails any idea?

[gemmellr]

I have updated the PR but can't see why the build fails any idea?

It seems to be unrelated, I see similar sporadically failures in a lot in other recent prior builds. Perhaps the test failure to look at before release @clebertsuconic (perhaps its even the thing you already noted looking at).

[laeubi]

@gemmellr @gtully I think this can be used as a first step, just one thing I noticed (but that could be changed later also):

Even though I removed the storage of multiple providers for the same scheme it might still be valuable to restore this. I noticed that e.g. the java sasl API do it the way that they (in case of multiple ones) ask them in order and return the first one returning a non-null client/server object.
We could do a similar thing here, MechanismFinder#getFactory would then simply return a list of factories.

[clebertsuconic]

nice PR... ran the whole test suite with this commit pulled and it worked fine

I squashed all 4 commits into one before I merged it.

[laeubi]

Thanks for all the reviews and of course the merge 👍

[jbertram]

Thanks much for the contribution, @laeubi.

[clebertsuconic]

yes.. thanks a lot @laeubi