It’s been less than a month since the last release of Pazpar2, but we are pumping new features into the software pretty quickly at the moment. For information about the most recent release before this, see here.
We have added a new method for normalizing MARC records (and OPAC records with embedded MARC data). We have found that we could achieve an 80% reduction of the CPU resources required to normalize bibliographic records by using a simplified XML representation instead of MARCXML. Because record normalization is a substantial part of the workload for Pazpar2, this enables us to handle twice as many users at a time, or search twice as many targets without sacrificing response time. The kit includes new stylesheets to process the format. MARCXML is still supported, if you already have stylesheets that you need to continue supporting.
The modified stylesheet for MARC has more sophisticated logic to determine the material type associated with records. It is a work in progress, but any feedback would be welcome.
New examples of service definitions are included in /etc/pazpar2/services-available (if you install Debian packages). Dynamically selectable service definitions are not for everyone, but it allows for very flexible applications of the technology, e.g. in SaaS environments.
Some bugs have been fixed in the brand-new threading support, and the functionality has been ported to Windows.