On Mon, 7 Jan 2013, Stuart Halloway wrote:

Problem found.  Implementation inheritance + java.lang.Cloneable for the
lose. :-)

Yes, that rings a bell for me. I have a Python implementation of a Lucene store and the cloning of IndexInput instances provides for some debugging fun and obfuscation :-)

Andi..


Sorry for the noise.




On Mon, Jan 7, 2013 at 5:23 PM, Michael McCandless <
luc...@mikemccandless.com> wrote:

On Mon, Jan 7, 2013 at 2:34 PM, Stuart Halloway
<stuart.hallo...@gmail.com> wrote:
Still no joy isolating, increasingly convinced the problem is in my code.
The overseeks are always in the .frq files, and always modest, e.g. 10%
past
the end, not 1000%.  Always from the ConjunctionScorer -- I can search
for
every word in the index in isolation without hitting the problem.

Strange ...

Any of that ring bells?

Sorry no bells ringing ...

When you run this same test on one of Lucene's stock directories does
it hit EOFE?  Ie, seeking beyond end of file and reading ought to
throw EOFE with the existing dir impls ...

(RAMInputStream seems to allow seeking beyond the end, but a
subsequent read should throw EOFE).

Mike McCandless

http://blog.mikemccandless.com

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org




---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org

Reply via email to