| Re: [jaxrs-dev] Vendor commitments |
Andy, thank you for this clarification! :-) Regarding TCK and spec I need to say that people do not care about claims of compliance, but about actual compliance. Everybody can go on Github and see what we decided. Everybody can read the Javadocs to learn about the new APIs and the optional bricks. No need to repeat the same text in the spec; least people read the spec. Everybody can download your software and simply try out whether it works or not, and see whether the optional aspects are supported or not. In fact, this is what most people actually do anyways, as passing the TCK never gave 100% certainty in the past (the TCK was not perfect and never will be). So my opinion is, what IBM does makes people a bit happy, but "really happy" is somewhat different and can only be reached by far earlier releases. Yes, IBM was among the first, and needed months, not years. But what people will expect from the new cadence is weeks, not months anymore. Anyways, good to hear that JAX-RS will be supported by upcoming releases of IBM software! -Markus From: jaxrs-dev-bounces@xxxxxxxxxxx [mailto:jaxrs-dev-bounces@xxxxxxxxxxx] On Behalf Of Andy McCright Yes, you have completely misunderstood my post... Firstly, I said that I agree with you that vendor implementation is essential. Hopefully you don't find this to be sad news. :) Second, I mentioned that I don't think it is wise for vendors to make commitments to implement a spec without a spec document or TCK. I thought this should have been fairly obvious, but I will attempt to explain further. I could claim that IBM already implements the Jakarta's JAX-RS 2.1.1 spec because it must be compatible with Oracle's JAX-RS 2.1 spec, which we already implement - and are certified on Oracle's web site. However, other than the transitive property (IBM implements X, X is supposed to be compatible with Y, so therefore IBM implements Y), what proof does any Jakarta user have that IBM really implements this spec? And if transitive property is enough, then why are we bothering with this release? Further, per our discussions in GitHub and the mail list, we have agreed that certain things (like the Java SE JAX-RS impl) would be considered optional, but without a spec document, how does a user know that? And without a TCK, what prevents an implementation from saying that they are 2.2 compatible right now? Perhaps I am being naive, but I would like to believe that products that claim compatibility with a spec have actually tested and verified that they are in fact compatible. I can say that IBM has rigorously tested WebSphere Liberty for Java EE compliance in the past, and my expectation is that we will continue to do that for Jakarta EE compatibility. Lastly, I said that I am hopeful that IBM will have a production-ready (GA) release ready within six months or sooner of a given spec release. Given that IBM provided one of the first (if not the first) implementation of Java EE 8 - and that was closer to 9 months after the EE8 spec release, I think that should make users happy. In the past, production-ready implementations of the EE specs were not generally available for a year after spec release. Again, I would think this would make you happy, not sad! :) To summarize: 1) Vendor involvement is important. 2) We (the community) either need to continue to press Oracle to release the spec doc and TCK - and find a suitable workaround. 3) IBM is invested in the success of JAX-RS and Jakarta EE. Hope this makes you happier! :) Thanks again, Andy On Wed, Aug 29, 2018 at 9:13 AM Markus KARG <markus@xxxxxxxxxxxxxxx> wrote:
|