[redland] Redland News 2000-02-21

[Dave Beckett]

Here's what I've been working on recently for Redland, post the 0.9.8
release.

* Thinking about other language interfaces, hence the message of 7th Feb.

* Fixed sirpac parser wrappers to handle multi-line literals (was a
  known bbug on todo list)

and the big one:

* Removed all static allocations, parameterised all classes to be
  dynamically allocated.

This allows multiple instances of Redland to be running in the same
memory space, which helps with threading and using Redland as a
shared library.  However, there are no locks yet on the data
structures so threads won't work yet, but at least all the stuff is
in one place now.

This did mean creating an environment class (called 'world') and
changing all the main object constructors to take that as a
parameter.  This meant lots of changes at the C interface but all of
these have been hidden from perl and python, just to make things easy
for the common case - 1 instance.

The Tcl interface, as a very direct flat interpretation of the C
interface, sees all of these changes unfortunately but on balance, I
think it was worth it.

I think this may also help pure OO interpretations of Redland to work
better such as Python's and maybe C++.


This was one of the pre-1.0 things I wanted to get done.  Others
might include:
   Adding arcs_in/arcs_out methods to the model
   Adding a convienience container class
   Getting Rapier to work a lot better!

To that end I've been working on a test system for RDF/XML parsers
(all the ones Redland supports), using RDF/XML and Redland to do it
of course.  More details in a few weeks.

On Sunday I'm going to the W3C RDF Interest Group meeting in Boston
along with some of my ILRT RDF colleagues and then onwards to the KT2001
conference in Austin to speak about, guess what - RDF tools.

Ciao

Dave