Firstly, let me say this is a very well-written email. In order
to better inform some of my own thoughts on this, I started a
Twitter poll:
https://twitter.com/reza_rahman/status/1124827840652939269. The
results of the poll are very clear thus far but it will end in
five days. In my experience, the polls are usually well stabilized
by around 500-1000 input points. So far I have 600+ input points
and here are the results:
* 21% - Keep as long as possible
* 64% - Ditch as soon as possible
* 13% - Not sure yet
* 02% - Other (please comment?)
I am happy to say this coincides with my own thoughts and the few
customers I have had an opportunity to speak to about this
personally. The thinking reflects the trade-offs already mentioned
here so there is no productive purpose in repeating them. I
therefore am voting for Proposal 1.
I urge that the Eclipse Foundation do a poll similar to mine -
perhaps using the text here as the poll opener. While people care
a great deal, they have busy lives. I suspect we won't get 600
response emails here. Hopefully the Eclipse Foundation will get
thousands of responses to a poll. If that is really not something
we wish to do, I would appreciate at least retweeting my tweet.
For me, there is unquestionable value in a decision like this
getting the broadest input possible.
Below are additional responses. I am removing extra text for
brevity.
On 5/7/2019 7:23 AM, David Blevins
wrote:
This is a tough one as specifications have transitive
dependencies. It is perhaps better to start with a list of
specifications that should move: WebSocket, JSON-B, JSON-P,
Servlet, JSF, Batch, Concurrency, CDI, Bean Validation, EJB (I am
a little sad I need to include this. I wish we had replaced it in
Java EE 8), JPA, JMS, JavaMail, JAX-RS, JAX-WS, Java EE Security,
JAXB.
Yes. Everything that does not make the transitive list of
dependencies above. We could go a step further and aim to remove
even more in the future. There is a lot of cruft.
Stick with Java SE 8 if possible. That is what a majority of
customers are still using. They have their reasons. Upgrading to
Java SE 11 for Jakarta EE 10 is a good idea. There is good stuff
that could be absorbed into Jakarta EE. In fact, we are not really
even done adopting Java SE 8 yet.