Thanks for the explanation, I'm not in this situation but it's helpful to understand better lucene.
Nicola Michael McCandless <luc...@mikemccandless.com> wrote: >On Tue, May 14, 2013 at 10:02 AM, Nicola Buso <nb...@ebi.ac.uk> wrote: > >> I now this can sound horrible/flexible/... but this mean I can add two >> documents with the same field name, but different configurations, for >> example different IndexOptions? > >Yes and no :) > >Lucene will happily index such drastically different fields, but then, >on merge, it will "merge" those two FieldInfos and pick the same >configuration for both. For example if one of the fields omitsNorms >and the other did not, then on merge, they both will omit norms. >Other properties about the field, e.g. whether it was indexed or >whether/how term vectors were stored, are not merged/changed because >this is a per-document thing. > >Lucene is often claimed to be "schema-less" but the truth is it >effectively has a secret schema that can change over time as merges >are done, if you index the same field name differently over time ... > >Mike McCandless > >http://blog.mikemccandless.com
