From: James M. <jma...@ps...> - 2003-02-26 22:18:00
|
At 09:35 PM 2/26/2003 +0100, Andrea Aime wrote: >Hi everybody, >I think that the new shapefile datasource is good enough to substitute >the old source base. The new module is derived both by the current >code base and a new code base that comes from the JCS (Java >Conflation Suite), in particular, the shape handling comes from JCS, >the dbf from the old (with improvement ported from the JCS code base), >the ShapeFileDataSource is completely new code. >I've tested it with a collection of 216 shapefiles I have on my disk, coming >from ESRI (it was included in the "GIS for everyone" book), and overall it's: >* generally faster (not always, but on the vast majority of files it is, > especially with some big files is twice as fast); >* handles more incorrect situations gracefully; >* can write shapefiles :-) This is excellent news. I seem to be getting new Shapefile related code from all sides at the moment so I think the best thing to do is to get your code into CVS so that everyone can see it and see where their code fits in with it. >Finally, the endian data sources would be probably better inserted into the >resources module I guess, since endianness may be a problem with >every file data source we have... One person has recently contacted me to say that they have a version of the shapefile module working using the nio package which allows for endien switching. I've not seen the code yet so I don't know how hard it will be to layer what he has done with what you have done, but the end result may be that we don't need the endian code. Anyway, that's further down the line, for now I'd like to see your code in the CVS, if possible could you include some of the sample files you have been using and tie them to Unit tests. Its probably best not to put in any that are too large as the unit tests should be there to test functionality not speed. Many thanks for your efforts James |