lucene-java-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From nel...@monkey.org (Nelson Minar)
Subject Field.Text(String, Reader) vs. Field.Text(String, String)
Date Fri, 28 Sep 2001 21:09:47 GMT
One strange "gotcha" in Lucene:
  Field.Text(String, String) creates a field that is stored, but
  Field.Text(String, Reader) creates a field that is *not* stored.

I'd naively assumed that the two Field.Text() methods were just for
convenience, I hadn't expected that the semantics would change
depending on what way I got the data into the Field. That seems like a
misfeature to me.


For my own purposes I did a little chart of all 8 possible field
types, and which factory methods create which. The docs are all
consistent with the code, just seeing it this way made it clearer to me.

Stored  Indexed  Tokenized
   yes      yes        yes     Field.Text(String, String, String)
   yes      yes         no     Field.Keyword(String, String)
   yes       no        yes
   yes       no         no     Field.UnIndexed(String, String)
    no      yes        yes     Field.Field(String, String) , Field.UnStored(String, String),Field.Text(String,
Reader)
    no      yes         no
    no       no        yes
    no       no         no

One thing that pops out is that it never makes sense to tokenize but
not index. Similarly, it never makes sense to neither store nor index.
This all seems obvious in retrospect, but it's helped me understand
how tokenizing, indexing, and storing all fit together.

                                                     nelson@monkey.org
.       .      .     .    .   .  . . http://www.media.mit.edu/~nelson/

Mime
View raw message