« Apple switching to Intel processors | Main | Two Towers »


Getting closer to releasing Wilbur2

One of the reasons for moving from Wilbur to Wilbur2 is not only that I wanted to do a redesign of (some parts of) the system, but that I also wanted to "reorganize" the source code, and with CVS this seems so cumbersome. So I have decided that the old Wilbur will remain in the CVS where it is, and I will start a new module altogether (perhaps called "wilbur2" or something).

At the moment, I have a prototype Wilbur2 running, with source code arranged in directories approximately as follows:


I separated "nox" into a directory of its own, mainly because it uses a separate package. The "goodies" directory contains stuff that (strictly speaking) is not needed for running Wilbur, but can be useful (also, some "legacy" stuff is there as well). I anticipate putting some contributed code there too. And yes, I have given up on trying to find the ultimate defsystem and decided to just go with ASDF.

The RDF schemata that comes with current Wilbur will be deprecated. Just load the stuff from where it is supposed to be loaded...

For those who are curious, "index-and-match.lisp" contains stuff that supports my RDF browser ("OINK", yet to be released), things like full-text indexing of literals, regexp matching, etc.

I also have a SQL-interface in the works, based on CL-SQL, but it will require more work to make into the first release of Wilbur2.

Posted by ora at 10:30


Of course, I'd suggest putting Wilbur2 into a darcs repository. This would be handy for all sorts of reasons -- firstly, this kind of reorganisation, and secondly it'd be easier for me to make changes and throw them back up to you :D

I'm sure the common-lisp.net people, or the tech.coop, would be willing to discuss this if Sourceforge is unforgiving.

Re naming and directories:

core/core-data.lisp: not core/data.lisp?

(similarly for *-package.lisp -- the convention as I've seen with ASDF is to call all package definitions defpackage.lisp, relying on the package and directory to disambiguate.)

Seeing 'wilbur.rdf', it strikes me that you could put a DOAP description in there... I've even got the RDF if you want it (just give me a URI for Wilbur itself) :)

All good!

Posted by: Richard Newman at June 7, 2005 12:35 PM

Thanks. These all seem like good suggestions.

Posted by: Ora Lassila at June 7, 2005 12:52 PM

So I guess the critical question is: What is the URL for Wilbur? Any suggestions? Perhaps "www.lassila.org/wilbur"...?

Posted by: Ora Lassila at June 7, 2005 12:54 PM

I'd probably suggest something future-proof:




that way, should you ever get the opportunity you can stick a URIQA server at /rdf on your host, and get CBD lookup for nothing.

I've mailed you some DOAP!

Posted by: Rich at June 7, 2005 02:02 PM

Subversion's pretty versatile when it comes to moving things around:


(I moved over to it last time I broke my csv, a definite improvement)

Posted by: Danny at June 8, 2005 07:07 AM

Warning: Totally off topic Hi Ora, i just found out about your blog. GREAT! I can now easily keep track on your work! Hope you are doing well and I promise this is the first and last comment on non-RDF matter :-) Greetings from Switzerland.

Posted by: Jerome at June 8, 2005 10:31 AM

Jerome: good to hear from you. Non-RDF comments are allowed ;-)

Posted by: Ora Lassila at June 9, 2005 01:18 PM

Sir, I would like to develop a semantic web using wilbur in my pg project.So, provide me with the deatils of wilburand how far it is helpful to me to developm a project on semantic web and so on. Please reply to email.

Posted by: S.NAGADEVI at July 30, 2005 12:57 AM