Release 1.0.3

Upgrading from 1.0.x

It is necessary to reindex the complete index in order to be able to retrieve the feature values via the fl parameter.

In coordination with The Apache Software Foundation the shipped jar file has a modified filename. Please delete your old pixolutionForSolr.jar and modify your classpath imports accordingly to the new name pixolutionForApacheSolr.jar.

New Features

Added support for fl parameter

The feature can now be retrieved as text representation via the fl parameter. This may be useful for exporting the features in an external database and importing the data via a DataImportHandler (DIH). This feature is also a step towards distributed search support.

Bug Fixes

Document not found with rank.by id

The ID from a rank.by query was searched in raw format instead of using the internal fieldtype specific format. If the fieldtype for the document ID was a numeric or processed field then rank.by ID searches wouldn’t work. This is fixed and the given ID is translated into the appropriate internal format before searching for the corresponding document.

Incorrect file too big exception

Several images can be loaded using the same socket connection. Their filesizes were then summed up and could lead to an incorrect file too big exception depending on the processor.loader.fileLimitkBytes setting. This is fixed.

Temporary image files not always deleted

When an error occured during indexing or querying by URL, the temporary image files (*.image) of the downloaded images may have not been deleted from disk. This could pollute disk space. These temporary files are now always deleted.

Other changes

Improved image loader error messages

When an image could not been loaded the error messages are now more meaningful. Especially for not supported protocols when a URL is given.

Modified jar file name

In coordination with The Apache Software Foundation, the delivered jar file ships now with a modified filename. It is now pixolutionForApacheSolr.jar.