Hello all
Rémi finished today to attach the patches for the upgrade to ISO
19115:2014 (metadata) [1], [2]. I'm in process of reviewing and
integrating the patches in both Apache SIS and GeoAPI. I'm doing this
work on the branches and plan to merge to trunk only when the work will
be fully completed, since the trunk will need a special treatment. This
is because the trunk uses the GeoAPI 3.0.0 release, which will not
contain the new interfaces. The new interfaces are currently available
only on GeoAPI snapshots, and will be proposed at OGC for a release
later. For the trunk, I plan to reference directly the Apache SIS
concrete class when the GeoAPI interface is not available.
Deprecated metadata methods may need to stay there for the next 10
years, since the European Union is not planning to upgrade their
metadata schema soon. I think that USA (e.g. NOAA) may be faster.
However while deprecated metadata methods are still there, the
corresponding private fields are removed. Instead, deprecated methods
delegate their work to the new methods, which will store the information
in their new private fields.
As a side-effect of metadata work, I provided an initial implementation
of org.opengis.util.Record [3]. Those records are similar to records in
the Fortran language and are used in some metadata elements and in
coverages (rasters). A "record" is very similar to a "feature" where all
members are constrained to the [1 ... 1] cardinality. This is also known
as a "simple feature". This cause a duplication of code and concept
between "Record" and "Feature", which I'm not sure how to handle... I
would prefer to handle "Record" and a special kind of "Feature", but
problem is that ISO standards define those two structures in quite
different way and use them in different places.
Martin
[1] https://jira.codehaus.org/browse/GEO-232
[2] https://issues.apache.org/jira/browse/SIS-94
[3] https://issues.apache.org/jira/browse/SIS-127
