MDweb 2010.06 Release ------------------------------------- The MDweb team is pleased to announce the 2010.06 Release of MDweb. All improvements are now finished. MDweb is distributed as a web archive file, which can be downloaded over the Internet from the directory: http://www.mdweb-project.org/files/releases/ as file: mdweb-2010.06-2.2.0.war The file can be deployed directly into a JEE application server such as Glassfish or Tomcat. MDweb will need a database running on a PostgreSQL database management system, and the postgres JDBC driver installed in the JEE application server. Documentation on the installation and use of MDweb can be found in two documents, a Quick Start Guide and a Users' Manual. The documents are avaliable online in the directory: http://www.mdweb-project.org/files/documentation/2010.06/ as files: Quick Start Guide: MDweb-QuickStartGuide-2010.06-en.pdf Users' Manual: MDweb-UserManual-2010.06-en.pdf in both French and English. This version is known to run on various JEE web application servers including Tomcat and Glassfish. Deployments of multiple instances of MDweb on the same machine are now supported but will only work for version 6 JEE servers (JEE6) and this has only been tested on Glassfish v3. This is the first release after a complete rewrite of MDweb from its PHP based origins to a Java based web application. This new version has not been extensively tested in the wild; we welcome reports of bugs or other issues in our public issue tracking system: http://jira.codehaus.org/browse/MDWEB on Codehaus. Certain aspects of the application, notably related to workflows and long term management of users and of metadata records, have not yet been fully conceptualized and designed. Similarly, the Metadata schemas and profiles have not yet been tested extensively and probably will warrant improvement. These are aspects that will be tacked in future versions of MDweb. Highlights of this release: -------------------------- * Platform: MDweb is now entirely based on the Java programming language. Each instance runs in a Java Enterprise Edition (JEE) application container. * Platform: Multiple MDweb instances can now be deployed on the same machine when run in a JEE 6 platform, restricted only to all having distinct application names. * Platform: MDweb can be used against either a PostgreSQL database backend or a local embedded database system (based on Derby). * Application suite: the suite of web applications can now be themed with a custom logo, favicon, title, and colours to a limited extent. * Application suite: MDweb has a configuration component through which to parameterize the applications. Configuration can now be done entirely through the web application itself without editing configuration files by hand. * Search: the search application can be used to consult multiple remote OGC Catalog Services for the Web (CSW) service, working concurrently and asynchronously. * Search: the search application can be configured with custom background maps and an initial zoom to a particular area of interest. * Search: individual metadata records can now be displayed directly in their own separate page by requesting the appropriate URL. * Edit: the metadata record editing application enables the creation and editing of metadata records directly from the base profiles or from sitewide or user templates. * Edit: metadata records can be imported from earlier versions of MDweb with a high probability of success, although some fields might be lost during import. The importing system automatically suggests a profile for import when it can recognize the XML file and informs the user when elements might be lost. * Admin: MDweb now has an administration component through which new users can be added, extra RecordSets can be created, and the Catalog's index can be refreshed. * Configuration: the initial configuration of MDweb can now be completed by the first user to connect to the application suite. This initial setup is limited to the strict minimum required to run the application; configuration can be completed through the configuration application. * Configuration: all configuration can now be done entirely through the web application itself without editing configuration files by hand. * Configuration: Some controlled vocabulary databases, notably for GEMET and Agrovoc, can be downloaded automatically and inserted as the vocabulary offered by the internal Thesaurus service. * Schemas: A base set of profiles based on Dublin Core, on ISO 19115, or on INSPIRE can now be used to create new metadata records. * Schemas: Experimental schemas for the French profile of ISO 19115 and for the upcoming NatureSDI project are also available. * Documentation: MDweb now has documentation which can be found online at http://www.mdweb-project.org/files/documentation/2010.06/ in both English and French, and in both PDF and ODT formats. Known issues in this release: ---------------------------- Metadata Schemas and Profiles: ----------------------------- * XML: Empty fields are created in the XML document for certain elements rather than merely omitting the element. => This is mostly innocuous but could be improved; we need to examine the situation. * ISO_19115: Dates are reported as milliseconds rather than the way they were defined by the user. => we are in discussion with our partner projects and at the Open Geospatial Consortium, possibly also within ISO during the revision of the 19115 standard, for a way to resolve our handling of dates. * INSPIRE: We fail to correctly handle the field declaring conformance with a published data standard. The technical guidelines for mapping INSPIRE records onto ISO are attempting to use an XML boolean field to indicate three different states: untested, tested and passed, or tested and failed. This causes great difficulty for us since the XML parsing tools are not built for that case. => we plan to seek guidance from the editors of the technical guide at the European Joint Research Centre. Platform: -------- * The referencing subsystem does not correctly use the created database. This is a bug in the extension mechanism of the Geotoolkit.org referencing system. The Geotoolkit.org referencing system automatically generates its own embedded Derby database in the geotoolkit.org/ configuration directory. However, the situation does not work when running multiple deployments of MDweb on the same machine or runnning mixed deployments of MDweb and other programs that use the Geotoolkit.org referencing system. Such systems will require manually placing a 'DataSource.properties' in the EPSG/ subdirectory of the Geotoolkit.org configuration directory to point the referencing system to a shared EPSG database running on a PostgreSQL server. => While awaiting the fix from the upstream project, we are considering providing a way to generate the required file automatically. Unfortunately, doing this by default could break the installation of certain users so the situation must be examined further. Web Applications: ---------------- * Search app does not correctly handle real projections. Currently the only 'projection' correctly supported is the Platte-carre. This needs extensive testing in the upstream MapFaces project to understand the issues, their causes, and possible resolutions. * Search app occasionally fails to display the correct map. This is often encountered on the initial zoom because the map pane fires off two WMS requests and displays the last one returned not the last request made. This may need to wait for a switch to JSF2. * Search app: The vocabulary offered by the Thesuarus is not optimal since the concepts offered are an unstructured, haphazard set of concepts from the vocabulary. => We need to migrate to a more dynamic system for getting words from the thesaurus service. * Edit app: import of INSPIRE records adds an empty keyword field to otherwise complete records, turning them into elementary complete only. => minor bug for which all fixes are complex. * Edit app: Cannot yet link metadata records to a series parent record. => We need to fully design around the possible workflows involved in creating metadata records for resources in series. * Edit app: Cannot yet add file attachments to a metadata record, such as a thumbnail (aka 'quicklook'). => We need to add the ability to upload the file, tie the reference to the metadata record, and provide the file to clients through the web service. Documentation: ------------- * Usage Guide: We plan to add a guide explaining core ideas of resource management and metadata authoring for new users of MDweb. => this work is already underway but the document has not yet reached a usable state. * Authors' Guide: we plan to compose a guide explaining the general design of Dublin Core, ISO 19115, and INSPIRE metadata records. Security: -------- * HTTPS login: login currently occurs over HTTP and is therefore inherently unsecure. => We plan to move towards a more robust, secure, and modern security system, possibly leveraging the access control mechanisms of the JEE platform, and possibly tying into broadly shared systems such as OpenID. * User authentication and authorization: this currently happens internally to the application against a database of users stored in the chosen database management system (DBMS). => We plan to make this more flexible, leveraging the access control mechanisms of the JEE platform, to enable alternative sources of authentication and authorization systems such as LDAP databases. Web Services: ------------ * Thesaurus: The thesaurus service has not yet undergone any formal evaluation in its design. Future iterations will undoubtedly change its methods and improve its operation. * MDweb service: The MDweb service for remote operation has been disabled for now due to lack of time to fully implement and test it. --The MDweb Project Team. June 14th 2010, Montpellier, France.