Hi Martin, This is great. I think my goal while you are sending all of these emails will be to start documenting things on the SIS wiki [1], and then figuring how somehow to integrate the different website efforts for SIS by Adam, Ross and others. I know Andrew was also working on a logo too.
In the meanwhile, keep the great rationale/documentation coming! Cheers, Chris On Oct 15, 2012, at 2:43 AM, Martin Desruisseaux wrote: > Hello all > > I just started some more "domain specific" commits (not yet merged to trunk). > First, there is an org.apache.sis.internal.simple package [1], which contains > trivial implementation of some GeoAPI objects (for now only Citation). The > OGC/ISO CI_Citation type is used very often in various corners of the > international standards. It is used for example in order to specify who > defined Coordinate Reference System codes. Because CI_Citation is a metadata > object, it will be defined in the sis-metadata module. However because > CI_Citation is used in so many corners, it will be needed in a few places of > sis-utility. So the org.apache.sis.internal.simple package in sis-utility > defines a trivial implementation where only the title attribute is stored; > all other attributes are null or empty. > > I will talk more about the role of CI_Citation in sis-utility later this > week, when I will bring the wider scope of how to manage the various kind of > identifiers in ISO 19139. (Note: ISO 19115 is the international standard that > defines metadata objects. The org.apache.sis.metadata.iso package will be > modelled by this standard. ISO 19139 specifies how to express ISO 19115 > objects in XML). > > Next, there is an initial commit of a org.apache.sis.xml package [2]. This > package is the proposed main entry point for XML marshalling/unmarshalling in > the Apache SIS library as a whole. I will talk more on this issue later, as I > introduce classes. For now the only class is Namespaces, which define static > String constants for namespaces frequently used in OGC/ISO standards. > > The proposed framework for XML (un)marchalling is based on JAXB for many > objects, but not all. The proposed approach is: > > * Use JAXB when the structure of objects is quite complex, but there > is not million of those objects (so performance is not critical). > Metadata typically fall in this category. > * Use SAX parser when the objects are relatively simple, but we may > potentially have million of such objects. Simple geometries like > points typically fall in this category. > > > Please let me know if you have any comment... > > > Martin > > [1] > https://svn.apache.org/repos/asf/sis/branches/JDK7/sis-utility/src/main/java/org/apache/sis/internal/simple/ > [2] > https://svn.apache.org/repos/asf/sis/branches/JDK7/sis-utility/src/main/java/org/apache/sis/xml/ >
