JakartaEE Release???

[hazendaz]

I see comments on netty 5 being part of version 3 but what about releasing as-is with jakartaEE? Selenium won't pick up the jakarta changes otherwise because of it being 'beta'. Seems to me making a cut that is jakarta and not waiting on netty 5 too when I see currently this sit still netty 4 would be a good thing. At least in our use case, we dropped the support we had for this through selenium as a result as we moved over to jakarta. We would like to restore this but seems some teams really need that final version before they jump over. Not sure that is even possible but would be a nice thing to have :)

[hyperxpro]

Sorry, I didn't get it clearly. Are you trying to integrate JakartaEE with AHC?

[hazendaz]

Sorry let me try to restate this better as its smaller issue than it sounded.

AHC 3 beta's moved to JakartaEE copy of jakarta activation. Technically EE9 copy since ACH didn't move over to angus-activation that replaced that (ie drop in place move) for EE10.

So for those of us using JakartaEE, we end up with two copies of jakartaEE activation or more specifically the EE 8 copy from AHC 2 and the EE 9 or 10 copy (EE10 angus-activation in our case).

Now jakarta.activation is probably one of few that doesn't need a container to start with so its not as huge a deal as other jakartaEE items are. But we still have a duplicate now and cannot simply exclude one. And we are looking to reduce library usage where possible given our stack is rather huge. This part adds to confusion.

The selenium team was asked to start using AHC 3 for this reason a while ago by someone else (don't have the link handy). But their response was that they would not pick up a BETA release per their policies.

So, from poking around here, it appears to me that AHC is waiting on netty 5 to be part of AHC 3.

So my ask was really to just cut something final that addressed the jakartaEE situation as its confusing enough assuming this is otherwise stable and then selenium may pick up for their selenium-remote-driver which in turn is a dependency for all their browser drivers.

Maybe that doesn't mean it has to be ACH 3, maybe 2.13.0 given I think this is a lower risk item in jakarta space. But wasn't entirely sure what all AHC 3 has in it now that would drastically change anything and that I guess would cause team to have to say version 4 for netty 5 stuff. So after rethinking, maybe a 2.13.0 cut with that change would be enough.

Anyway, really trying to see how we get a release out here that is on angus-activation 2.0.1 so it aligns better and then try again getting Selenium to pick that up as it would be final.

For reference in our case, we have a selenium based cucumber java stack which is supporting 100 different projects. We just switched over to jakartaEE and while we are working in this specific aspect as I noted, we have both jakarta.activation 1.2.2 from ACH 2 and angus-activation 2.0.1 coming from the bulk of other jakartaEE portions in use. So we would like to at least cut that part out and need some help from ACH to make that happen.

After saying all that I can see how then the flip could be a problem. Jakarta import split has made everything so complicated.

If this sounds like something feasible for 2.13.0, I'd be more than happen to send a PR on it, assuming to the 2.12.4 snapshot branch.

[hazendaz]

FWIW - its working as it is for us after we flipped over to JakartaEE so its really a minor library set situation and only not an issue because we used angus-activation. For any that would try the same and for whatever reason didn't know about angus and were trying to use jakarta.activation, they would have a problem.

[alexandernajafi]

@hazendaz I think I'm one of thos who don't know about Angus. Could you please elaborate what you mean? I get errors like

Caused by: java.lang.ArrayStoreException: arraycopy: element type mismatch: can not cast one of the elements of java.lang.Object[] to the type of the destination array, jakarta.activation.MimeTypeRegistry