[
https://issues.apache.org/jira/browse/JENA-2289?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17497342#comment-17497342
]
Lorenz Bühmann commented on JENA-2289:
--------------------------------------
Sure: https://github.com/apache/jena/pull/1204
Let me now if things have to be changed. The GeoSPARQL code contains too many
paths and methods, I might have missed something or even added duplicate code.
> Geospatial index not being loaded with assembler
> ------------------------------------------------
>
> Key: JENA-2289
> URL: https://issues.apache.org/jira/browse/JENA-2289
> Project: Apache Jena
> Issue Type: Bug
> Components: GeoSPARQL
> Affects Versions: Jena 4.4.0
> Reporter: Lorenz Bühmann
> Priority: Major
>
> Using the GeoSPARQL component via assembler doesn't load the geospatial index
> from disk and therefore doesn't register it to the context of the dataset
> which makes querying fail on a second startup.
> In the method {{prepareSpatialExtension}} in {{GeoAssembler}} class we have
> currently the following cases handled:
> case 1: no file given
> case 2: file given and either doesn't exists yet or empty
> Thus, case 3 is missing:
> case 3: file given and it isn't empty, i.e. the index has been precomputed
> First I thought just ignore the file existence/emptyness check and go with
>
> {code:java}
> GeoSPARQLConfig.setupSpatialIndex(dataset, spatialIndexPath.toFile());{code}
>
> But this would lead to an unnecessary retrieval of the SRS occurrences.
> What I did for testing is to add
> {code:java}
> SpatialIndex si = SpatialIndex.load(spatialIndexPath.toFile());
> SpatialIndex.setSpatialIndex(dataset, si); {code}
> but maybe we want to have this case also encapsulated in the
> {{GeoSPARQLConfig}} class? Something like
> {code:java}
> GeoSPARQLConfig.setupFromPrecomputedIndex(dataset,
> spatialIndexPath.toFile());{code}
> maybe?
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
