5 Model Generalization
99
For the WFS client-side software there are a number of options. WFS clients can
either be
• HTML-based Web clients that can be run in an Internet browser like Mozilla or
Internet Explorer;
• desktop applications that have to be installed first on the user’s computer. Also
in this case there is a choice between open source clients (uDig, Gaia) and WFSenabled desktop GIS or CAD software of companies like Intergraph (GeoMedia); or
• other Web services, for example a WMS service that plays the role of Portrayal
service for the GML output of the WFS service (see Fig. 5.11).
In this three-tier architecture the geo-database software or data-storage formats
that are used in the data layer are not ‘visible’ for the WFS service–client interaction.
The WFS client does not have to now the implementation details of the data layer,
because it only communicates with the WFS service. So in the next section, we will
concentrate on the WFS service–client interaction.
5.5 Server–Client Set-up and Progressive Refinement
Progressive refinement of data being received by a client could be implemented in the
following way. The server starts by sending the most important nodes in tGAP structure (including top levels of associated edge BLG-trees) in a certain search rectangle.
The client builds a partial copy of the data in tGAP-structure, which can then be used
to display the coarse impression of the data. Every (x) second(s) this structure is displayed and the polygons are shown at the then available level of detail. The server
keeps on sending more data, and the tGAP-structure at the client side is growing (and
the next time it is displayed with more detail). Several stop criteria can be imagined:
(a) 1000 objects (meaningful information density), (b) required importance level is
reached (with associated error tolerance value), or (c) the user interrupts the client.
In this section, we will look in some detail how the tGAP structure can be used
in a Web service/client setting for progressive transfer and for refinement during
zooming. As shown in Fig. 5.11, the WFS service acts as middle layer between Web
client and geo-database. So we have to establish two things: (1) what queries are
necessary from WFS to database to retrieve the vario-scale data, enabling progressive
transfer and smooth zooming later on (Sect. 5.5.1), and (2) what does this mean for
the requests and responses between Web client and WFS service (Sect. 5.5.2). We
will start with these two questions. In the last paragraph of Sect. 5.5.3, we will shortly
discuss the requirements of progressive transfer and refinement for the software used
for geo-database, WFS server, and WFS client.
5.5.1 Queries to the Database
From a functional perspective there are two situations in the Web service–client interaction: (1) the initial request to the WFS to get features for the first time during
the session; and (2) additional requests when zooming in, zooming out, and panning.
99
For the WFS client-side software there are a number of options. WFS clients can
either be
• HTML-based Web clients that can be run in an Internet browser like Mozilla or
Internet Explorer;
• desktop applications that have to be installed first on the user’s computer. Also
in this case there is a choice between open source clients (uDig, Gaia) and WFSenabled desktop GIS or CAD software of companies like Intergraph (GeoMedia); or
• other Web services, for example a WMS service that plays the role of Portrayal
service for the GML output of the WFS service (see Fig. 5.11).
In this three-tier architecture the geo-database software or data-storage formats
that are used in the data layer are not ‘visible’ for the WFS service–client interaction.
The WFS client does not have to now the implementation details of the data layer,
because it only communicates with the WFS service. So in the next section, we will
concentrate on the WFS service–client interaction.
5.5 Server–Client Set-up and Progressive Refinement
Progressive refinement of data being received by a client could be implemented in the
following way. The server starts by sending the most important nodes in tGAP structure (including top levels of associated edge BLG-trees) in a certain search rectangle.
The client builds a partial copy of the data in tGAP-structure, which can then be used
to display the coarse impression of the data. Every (x) second(s) this structure is displayed and the polygons are shown at the then available level of detail. The server
keeps on sending more data, and the tGAP-structure at the client side is growing (and
the next time it is displayed with more detail). Several stop criteria can be imagined:
(a) 1000 objects (meaningful information density), (b) required importance level is
reached (with associated error tolerance value), or (c) the user interrupts the client.
In this section, we will look in some detail how the tGAP structure can be used
in a Web service/client setting for progressive transfer and for refinement during
zooming. As shown in Fig. 5.11, the WFS service acts as middle layer between Web
client and geo-database. So we have to establish two things: (1) what queries are
necessary from WFS to database to retrieve the vario-scale data, enabling progressive
transfer and smooth zooming later on (Sect. 5.5.1), and (2) what does this mean for
the requests and responses between Web client and WFS service (Sect. 5.5.2). We
will start with these two questions. In the last paragraph of Sect. 5.5.3, we will shortly
discuss the requirements of progressive transfer and refinement for the software used
for geo-database, WFS server, and WFS client.
5.5.1 Queries to the Database
From a functional perspective there are two situations in the Web service–client interaction: (1) the initial request to the WFS to get features for the first time during
the session; and (2) additional requests when zooming in, zooming out, and panning.