Fast Infoset (Standard, GF project) can use an External Vocabulary for when two parties share vocabulary information. A practical approach to create such a vocabulary is to use XML Schema information, and Paul just created FastInfosetUtilities to do that. The results so far are very Promising.

Also check the discussion in the USER mailing list.

This is one of a series of blogs where I will try to highlight participation in several communities in GlassFish, specially non-Sun participation; perhaps some of these blogs will interest you as a user, or as a direct contributor.

The Fast Infoset project (some blogs [1] and [2], and white papers [3] and [4]) had wide industry participation in the standard. A number of people participated in the implementation including Alan, Ias, Don and the mailing lists for FI are active, both the DEV list and USERS list, with contributions from additional people like Tatu Saloranta (WoodStox) Jimmy Zhang (XimpleWare), Brian Potarelli (Inversoft) and Mark Swanson mark@ScheduleWorld.com. Ias has incorporated FI into FIME and I know he was also considering adding it into other places, including Mirae @ ASF, and zell@zell.best.vwh.ne reported work on integrating FI with XMLBeans. FI uses Japex which will be used as the testing framework in the EXI group, where Paul and Santiago continue to contribute.

Two words on the image of this blog. I've always liked Castells in part because they are build by a large group and everybody has a role, from very strong people at the bottom to very light boys - and girls - at the top. Even the crowd helps stabilize the castle. A great image for community building, which is why I used them when I was the lead for the JSP specs.

The FastInfoset project is part of GlassFish and provides an implementation of ITU-T Rec. X.891 | ISO/IEC 24824-1 (yep, a double standard spec!). It is a good implementation and comes with both StAX and SAX bindings, supports embedded binary data, and it has been used in other projects like in XJ3D to manipulate X3D and VRML documents.

Paul recently wrote about how nice it is to solve a customer problem:

The performance was pretty horrible though and I was *just* about to write a binary encoded XML mechanism when I came across your Fast Infoset stuff - which did *exactly* what I needed and saved a whole bunch of development and testing. It worked first time too!

In a related blog, Paul also explains the difference between Fast WebServices and Fast InfoSet. You may also want to check the Binary XML Forum and the USERS mailing list for FI.

Santiago has been working for a while on performance issues around Web Services and XML. He is a member of the original team that worked on the Fast Infoset standard and implementation (see the fi project), and is now contributing to the improvements to the iStack (JAXB 2.0 and JAX-WS 2.0) implementation in GlassFish. In his new blog he discusses the performance of JAXB vs DOM using results from the japex framework.