Skip to content
  • Daniel Gregoire's avatar
    Record and skip documents that fail to index in Lucene · d3cce8bc
    Daniel Gregoire authored
    Since we allow arbitrary code execution in the generation of the fields
    for document indexing, there are cases where ::meta/doc values call
    search when the search index is empty.
    
    There are an infinite number of other cases in which userland code could
    throw an exception during indexing.
    
    Rather than have the entire search index be empty if one document throws
    an exception, these changes print to stderr and record in an atom which
    imetas failed to index. The message printed to stderr suggests a remedy
    for indexing the specific imetas that failed.
    d3cce8bc