Thanks Tiago and all for the feed-back. You made me to rething the ways I could move forward on this front. The plan I'm now considering is: - Release a 1.0 version by the end of the year as initially planned - Quickly make the JSR proposal to standardize a 2.0 version of the API - Create two separate branches of the code base, one "1.x" branch and another "2.0" to follow the work of the expert group - Prepare users for a significant migration from 1.x to 2.0 as I don't want to restrict the liberty of design of the Restlet 2.0 expert group.
Advantages: - Answer to current users' desire to have a stable 1.0 release sooner than later - Allow a quick submission of the JSR, before the 1.0 release - Allow to sync the JSR with the next major JEE design cycle in case there is some synergy or integration possible - Leave the expert group free to redesign the API as necessary without backward compatibility constraints - Allow the expert group to keep getting some real life feed-back from existing Restlet users What do you think? Thanks, Jerome > -----Message d'origine----- > De : news [mailto:[EMAIL PROTECTED] De la part de Tiago Silveira > Envoyé : mercredi 26 juillet 2006 21:26 > À : [email protected] > Objet : Re: JSR and updated road map > > I have observed (from a safe distance) the work on Beanshell > and Groovy, > and I can attest to the old saying: code talks, and talk isn't code. > > What if the JCP expert group works on Restlets 2.0? What if, for a > change, an API is released for learning and teaching, collecting > feedback from real users in real situations, so that a new > API emerges? > > I am in favor of code first, specs second. When Hibernate > went from 1.x > to 2.0, people migrated, and it was sometimes a lot of work. > There were > tools to help and much documentation on the migration. There > are still > reminiscences of Hibernate 1 (the org.hibernate.classic.Session > interface, for example) in Hibernate 3, but the two are pretty much > incompatible. > > Same happened to Groovy when they changed the parser. For a > while, both > parsers were available, and the compiler was issuing warnings. > Eventually, the old syntax faded. > > I wouldn't say that the Restlet API has reached production maturity, > given the rate of refactorings that happened since beta 13. But I am > pretty sure that it will, long before the JSR is ready. Why wait? > > All the best, > Tiago > > Jerome Louvel wrote: > > Hi Lars, > > > >> I hope that the development is not stalled because of the JSR > >> approval process. > > > > My intent, as indicated in the road map, is to continue the > development > > normally, at least until the JSR is approved. Beyond that > point, I need to > > have a look at the JCP rules so see when the expert group > actually takes > > over the API design. > > > > Concerning the NRE, it will be the reference > implementation, but I do plan > > to keep improving it, fixing bugs, adding features during > the whole process, > > ideally staying in sync with the JSR work. > > > > Best regards, > > Jerome > > > >
