Hi,
I've been looking at doing some refactoring of jena-text code while
working on the Lucene upgrade (JENA-1250). The way query strings are
constructed in TextQueryPF and then just passed as potentially long and
complex strings to TextIndexLucene/TextIndexSolr is a bit awkward for
the new implementation of language-specific indexing that I did for
JENA-1250, which has to choose the targeted field names dynamically. The
old implementation didn't have this issue but it relied on a feature in
the Lucene 4 API that was dropped in Lucene 5 because it was considered
dangerous (for good reason).
I'm a bit stuck with the refactoring as I'd need to do some non-trivial
changes in the TextIndex interface that is implemented by both
TextIndexLucene and TextIndexSolr. I'm hesitant to change TextIndexSolr
because I don't want to break it further.
TextIndexSolr has suffered some bitrot and I'm not sure if it can be
used anymore. I remember trying to set up a Solr index for use with
jena-text but couldn't get it to work. There is very little
documentation about the Solr side of jena-text and no working unit tests
(there is a test class TestDatasetWithEmbeddedSolrTextIndex but it is
disabled with the comment "Embedded solr not supported").
So I'm left wondering whether the Solr support in jena-text serves a
useful purpose or not. Is anyone using it? I think it should either be
improved (starting from the documentation) or dropped, because currently
it's just a burden on the other code within jena-text.
Opinions?
-Osma
--
Osma Suominen
D.Sc. (Tech), Information Systems Specialist
National Library of Finland
P.O. Box 26 (Kaikukatu 4)
00014 HELSINGIN YLIOPISTO
Tel. +358 50 3199529
[email protected]
http://www.nationallibrary.fi
