In preparation for a 1.0 release, the source code has been moved to SVN. This will allow for better refactoring with branching in the case of the main engine and make it easier to keep the adapters in line with released RDF server versions.
Just to keep everyone informed and on the same page, I am currently putting the finishing touches on a major SPARQL upgrade. This will include several logic fixes, major ordering optimizations, and new language features in addition to a Jena adapter. I would guess this will probably be released within the next week.
The first complete SPARQL engine release was completed today. That means that it successfully passed all required test cases and should be now useful for general purpose functionality. While there are still some internal efficiency problems and there is only an adapter for Sesame currently, the release should be useful for evaluating SPARQL against an RDF server. For documentation and core code, download the main distribution "sparql". To use the the language with an RDF store, download the "sparql-sesame" release, which includes the command line tool and the JUnit tests against the specification test cases.
While still not finished by any means, the first file release of SPARQL is complete. Sorry for the large size (2MB) - I won't include the javadocs next time. There are a number of known issues included in the release notes, along with instructions on how to use an included driver class to test the distribution against an included Sesame 2 version. Now that the first file release is complete, there will be quicker releases from here on out.
The first draft of the design specification for SPARQL has been released at http://sparql.sourceforge.net/design.html, giving an overview of the project design. The design has major goals of server generality and deployability while supporting code extensions to increase specific efficiency.