jeudi sept. 17, 2009

Thoughts on REST-\*

OK, I really need more than 140 characters for this one (plus it's been a while since I blogged opinions).
Of course, this is personal opinion, not my company's.

Many people have reacted to JBoss' solo launch of REST-\*. What I'm concerned about is the approach, not the technology and specifications (I'm probably not the best person to comment on that part).

For one, having redirect to a web site is a "bad thing" (tm). JBoss being the only participant is also not giving the underlying technical effort a lot of chance to become a commonly accepted standard, and that's a pity.

Over on twitter, my colleague Jason sent me this sensible analysis which also questions the approach taken while finding some merits to the technical parts.

Mighty Roy is simply bashing and swinging at the proposal calling for . But then not many people get his blessing from day one.

Rickard Oberg is pointing out that JBoss has a interesting track record in terms driving the project (too bad the last paragraph on his affiliation to JBoss is actually taking some credit off of that assertion).

Contrary to what Haikal says, JBoss is not late to REST with their very decent JAX-RS implementation and their participation in the standardization effort. In fact, I'm not convinced by the SpringSource excuse (SpringMVC legacy) for not implementing this API.

Getting beyond all this criticism, it's rare enough to have people offer to do actual work to demolish it like Anne Thomas is doing. I've been with Sun for too long to throw away the baby with the bath water..

If someone wants to contribute standards, OASIS, W3C, or IETF is where it should happen. Granted you'll be better off starting from some specification or even better yet from a successful implementation (and you may end up not being able to call it REST-anything), but declaring REST-\* to the world and making it a one company thing sounds like a marketing mistake to me.

vendredi juin 19, 2009

GlassFish v3 a la carte screencast - Part 3 - Jersey and EJBs

This blog has moved to
Follow the link for the most up-to-date version of this blog entry.

In the first screencast, I installed a minimal GlassFish v3 from a small bootstrap (IPS toolkit), created a domain and started the server. The second entry did something actually useful with GlassFish and two containers: Java Web and Spring. In this screencast, I layer a custom distribution on top of a GlassFish kernel. Enough to deploy a JAXR-RS / EJB 3.1 (lite) application.

For the sake of brevity this screencast is mostly command-line. It starts with the 5MB ips bootstrap and installs a pre-defined custom distribution which is enough to deploy the jersey-ejb sample application. The custom distribution is essentially an IPS package with no artifact, only a set of dependencies on other packages. For the curious out there, here is the step-by-step for the screencast :

bin/pkg set-publisher -P --enable -O
bin/pkg set-publisher --enable -O http://localhost:10001 localRepo
bin/pkg install sample-distro
bin/asadmin create-domain --instanceport 8080 --adminport 4848 mydomain
bin/asadmin start-domain
bin/asadmin deploy ~/jersey/jersey/samples/jersey-ejb/target/jersey-ejb.war
open http://localhost:8080/jersey-ejb/

I hope this series of screencasts demystifies the IPS/packaging side of GlassFish and shows the interesting possibilities it offers to end-users.

mercredi avr. 18, 2007

Is Restlets the new Hibernate?

Jérôme Louvel has released Restlets 1.0. Congratulations to him and the team! Any 1.0 version is always something very special.

In this InfoQ interview, Jérôme is drawing a parallel between JPA and the new "Rest" JSR 311 (of which he's an expert member). Of course he would like for his Restlets implementation to be the main inspiration for the expert group just like Hibernate seems to have been for JPA. So Jérôme is not just a good developer and architect, he's also sounds like a good marketeer :)

In other comments on the article, I wish InfoQ's Stefan Tilkov would add GlassFish's JAX-WS to the mix when comparing Restlets to SOAP stacks. While the answer from Jérôme would likely be the same ("it's not really REST"), JAX-WS has Dispatch/Provider APIs to work at the lower messaging layer to implement any protocol including RESTful web services.

The other interesting part is the use of GlassFish Grizzly's technology in the Restlet framework as well as the use of the dual CDDL/GPLv2 licensing model which happens to be the exact same as what GlassFish uses... All this and the future promised WAR packaging could make Noelios (the company behind Restlets) a good GlassFish partner.

On a personal note, having only been thru the basic Restlet samples, I can claim that what I like best about the framework is its website. Clean and simple.

jeudi févr. 15, 2007

JSR 311 - Quick follow-up thoughts

- You can do RESTful applications today with Java, only APIs like Servlets or the dynamic part of the JAX-WS APIs aren't the most natural approach (of course you can also simply use the Apache HTTPClient or classes, but Java developers deserve better than that).

- The target is J2SE 5.0 or higher and Java EE 5 or higher. Maybe Java ME. And the draft/early/out-of-the-blue roadmap mentions a final release in 2008 (so not related to Java SE 7 or Java EE 6).

- The Reference Implementation source code for this JSR will be made available under the same terms as Project GlassFish (this is the Specification Leader's responsability).

- RESTafarian Jerome Louvel of Restlets fame is part of the Expert Group.

- Marc Hadley (Spec Leader and WADL creator), Henry Story, Eliotte Rusty Harold, Dave Johnson, Tim Bray and others are all picking up the news and voicing someconcerns. The Expert Group is still recruiting. Hop on guys!

mercredi févr. 14, 2007

What if we took the best of servlets and JAX-WS and made RESTful Web Services a first class Java citizen? Welcome to JSR 311.


This blog has moved

Alexis Moussine-Pouchkine's Weblog

GlassFish - Stay Connected


« juin 2016

No bookmarks in folder


No bookmarks in folder