Download presentation
Presentation is loading. Please wait.
Published byYuliani Setiabudi Modified over 6 years ago
1
Updating GML datasets S-100 WG TSM5 19-21 September 2017
Raphael Malyankar, Consultant, Portolan Sciences Eivind Mong, Consultant Work performed under NOAA sponsorship
2
Introduction The GML specification does not specify any particular format for updates to GML data files. Drafts of S-122 and S-123 define a ‘whole-object- replacement’ approach to updating datasets. At NIPWG 4 in May 2017, we were requested to write up the S-122/S-123 approach for the S-100 WG.
3
Options for updating GML datasets
Attribute-level update – update delta dataset describes updates to attributes as well as instances. Object-level update – update delta dataset describes updates only at the instance level. Plain text update: Update with a text-based ‘patch’ utility – update is a purely text-based delta between the current and updated dataset that ignores structure and semantics. XML-aware patching – update is an XML-based delta that identifies the XML item (attribute or instance) being updated. Whole-dataset replacement – the entire dataset is distributed when an update is necessary. XML-aware patching example for an addition of a restriction attribute to a RestrictedAreaNavigational feature: <add = ‘RESARE55’]"> <restriction>entry prohibited</restriction> </add>
4
Summarized Analysis – 1 Factor/Approach Attribute-level Object-level
Text patch XML patch Whole-dataset Update detail Attribute, instance, spatial object, spatial primitive Instance, spatial object Line-based Attribute, instance, spatial object, spatial primitive Dataset as a whole Schema Modified Same as base dataset Not needed Special common XML schema Robustness Flexible More flexible Least flexible Like Attribute-level. Robust Volume of update delta Low 10 X attribute level? Comparable to Attribute level Comparable to Attribute-level 100 X attribute level? Tools - availability OEMs must develop. Off-the-shelf plus custom ‘wrappers’ Uncertain. Open source tools exist, customization may be needed. No special functionality Implementation effort Highest Moderate Application logic Application logic like 8211 format. Simpler parts of 8211 logic None for basic implementation. Error recovery TBD Simple logic for basic implementation. Error recovery TBD Simplest Sotware re-use Reusable Re-usable
5
Summarized Analysis - 2 Factor/Approach Attribute-level Object-level Text patch XML patch Whole-dataset Update tracking and object history on end- user system More complex Simplest Most complex Comparable to Attribute-level Difficult unless a ‘dataset’ contains only a single feature. Does updating spatial type need special treatment? Whole primitive; updating single coordinate more complex As object; update to single coordinate more complex Depends on text structure As object; update to single coordinate more complex No special treatment Previous version requirements – producer & end-user Minimal Previous version needed Not needed Transfer format File Compatible with web services Compatible with web services Not compatible with web services Low compatibility with web services. As for base dataset Data product types Vector and/or info type Vector and/or info type Vector and/or info type and imagery, etc. Compatibility with GML 3.2.1 Possible Yes Yes - ignores GML altogether. Yes – update format is non-GML can reference items in any GML file. A more detailed version of this table is included in the paper.
6
Recommended methods Object-level updating for products based on discrete features or information types, with two caveats: Products where datasets are very small will generally need only whole-dataset methods; and Data or coordinate collections or sequences will need special treatment (e.g., functionality extensions to locate coordinates by their positions in the sequence) in order to reduce the average data volume of updates. Whole-dataset updating for imagery, gridded data, etc., with the caveat below: Methods for updating parts of gridded or pointset data in other formats should be adapted to GML encodings. Specific products may have exceptional circumstances
7
Details Specific details for adding whole-object update formats to a product specification are described in the paper (pages 5-10). For example: Feature ID and gml:id of updated feature must be the same as the original. Versioning of features – add equivalent of RVER in S-100 GML profile. Deletions – use fixedDateRange attribute. Loading requires all previous updates to be available (or applied). Cancelling datasets – update with edition # = 0. Update cannot change data coverage. Validation checks for (1) updates in isolation and (2) datasets after updating. Metadata for update datasets.
8
Conclusion The ‘whole-object’ approach is a reasonable compromise between the various factors for most discrete-feature vector datasets. Specific data products might have special circumstances which dictate a different approach. Suggested next steps: Evaluate the most promising approaches. Develop specifications, rules, or guidelines about update dataset format, processing, and management, for eventual incorporation into S-100. Develop supporting guidance for authors of product specifications and implementers.
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.