Warn if no pidField configured
Closed this issue · 1 comments
jan-niestadt commented
If you don't configure corpusConfig.specialFields.pidField
in your .blf.yaml
file, Lucene document ids will be used instead. But these are not persistent! Reindexing your data might assign completely different Lucene document ids, and if you relied on the document ids staying the same, you're now hosed.
Because it's easy to overlook this fact, IndexTool and BLS should refuse to index (with explanation) unless you configure a pidField
, or you explicitly set an option like assignNonPersistentDocIds
.
jan-niestadt commented
We'll probably make it a warning first.