Changes between Version 6 and Version 7 of FDORfc38


Ignore:
Timestamp:
07/16/09 06:28:07 (15 years ago)
Author:
oresthalustchak
Comment:

--

Legend:

Unmodified
Added
Removed
Modified

  • FDORfc38

    v6 v7  
    2323== Overview ==
    2424
    25 This RFC is for adding the FDO SHP provider capability to support multi-polygons geometries.
     25This RFC is for adding the FDO SHP provider capability to support multi-polygons geometries in the cases where SHP files have multi-polygons.
    2626
    2727== Motivation ==
    2828
    29 Currently, this provider supports only simple polygons. The SHP specification defines only a polygon type, but it defines it as supporting multiple outer polygon loops, which basically are multi-polygons as defined in FDO. The FDO SHP provider is not taking advantage of this and is not returning multi-polygons when the polygons are actual multi-polygons. The main processing that is missing with the current provider is on read, it does not check the polygon details to determine whether to generate an FDO polygon or an FDO multi-polygon geometry.
     29Currently, this provider supports only simple polygons. The SHP specification defines only a polygon type, but it defines it as supporting multiple outer polygon loops, which basically are multi-polygons as defined in FDO. So a SHP polygon can be either an FDO polygon or an FDO multi-polygon, depending on the number of outer loops. The FDO SHP provider is not taking advantage of this and is not returning multi-polygons when the polygons are actual multi-polygons. The main processing that is missing with the current provider is on read, it does not check the polygon details to determine whether to generate an FDO polygon or an FDO multi-polygon geometry.
    3030
    3131== Proposed Solution ==
    … …  
    4040== Implications ==
    4141
    42     * The code for this is mostly done, just turned off. We will need to measure any performance impact, but the extra processing would only apply for cases where there are multiple loops. The most common case of a single loop polygon would not be processed any differently and would not have a performance impact.
     42    * The code for this is mostly done, just turned off. We will need to measure and work on any performance impact, but the extra processing would only apply for cases where there are multiple loops. The most common case of a single loop polygon would not be processed any differently and would not have a performance impact.
    4343
    4444    * The FDO geometry type will be set accordingly on reading. The caller must expect either a polygon or a multi-polygon regardless of original geometry type on insert.
    … …  
    4747
    4848    * Enhance the unit test and add roundtripping fidelity tests.
    49     * Performance benchmarks on large datasets containing polygons with many interior rings.
     49    * Performance benchmarks on large datasets containing polygons with many rings.
    5050
    5151== Funding/Resources ==