lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Michael McCandless <luc...@mikemccandless.com>
Subject Re: A report for "Not so important (but still is a) bug"
Date Sun, 08 Nov 2009 17:46:08 GMT
I sure hope not :)  I hope 3.0 is out soonish...

But just in case, I've reopened & marked fix version 2.9.2, so if we
ever do a 2.9.2 that'll remind me to backport.

Mike

On Sun, Nov 8, 2009 at 11:18 AM, Simon Willnauer
<simon.willnauer@googlemail.com> wrote:
> dude, will we have 2.9.2? :)
>
> simon
>
> On Sun, Nov 8, 2009 at 4:14 PM, Michael McCandless
> <lucene@mikemccandless.com> wrote:
>> We should fix this; I'll open an issue & dig.
>>
>> Somehow SegmentInfo.files() is being buggy, claiming _0.prx is a file
>> belonging to the segment, when it clearly isn't.
>>
>> BTW we can't hold 2.9.1, since it's now "out" (as of yesterday) :)
>>
>> Mike
>>
>> On Sun, Nov 8, 2009 at 8:58 AM, Shai Erera <serera@gmail.com> wrote:
>>> Hi
>>>
>>> I wanted to check something and wrote this very simple program, which
>>> surprisingly failed because of a FileNotFound exception:
>>>
>>>       Directory dir = new RAMDirectory();
>>>       IndexWriter writer = new IndexWriter(dir, new SimpleAnalyzer(),
>>> MaxFieldLength.UNLIMITED);
>>>       writer.setInfoStream(System.out);
>>>       writer.addDocument(new Document());
>>>       writer.commit();
>>>       writer.close();
>>>
>>> It fails with the exception:
>>>
>>> Exception in thread "main" java.io.FileNotFoundException: _0.prx
>>>     at
>>> org.apache.lucene.store.RAMDirectory.fileLength(RAMDirectory.java:149)
>>>     at
>>> org.apache.lucene.index.DocumentsWriter.segmentSize(DocumentsWriter.java:1150)
>>>     at
>>> org.apache.lucene.index.DocumentsWriter.flush(DocumentsWriter.java:587)
>>>     at
>>> org.apache.lucene.index.IndexWriter.doFlushInternal(IndexWriter.java:3572)
>>>     at org.apache.lucene.index.IndexWriter.doFlush(IndexWriter.java:3483)
>>>     at org.apache.lucene.index.IndexWriter.flush(IndexWriter.java:3474)
>>>     at
>>> org.apache.lucene.index.IndexWriter.addDocument(IndexWriter.java:1940)
>>>     at
>>> org.apache.lucene.index.IndexWriter.addDocument(IndexWriter.java:1894)
>>>
>>> This happens on the latest code from trunk (and validated also against 2.4).
>>> Seems like it's there forever. This only happens when I set infoStream,
>>> because as part of the logging we compute the segment size. _0.prx is not
>>> found because I didn't add any terms to the index. If I don't set the
>>> infoStream, or add terms to the index, this exception is not thrown.
>>>
>>> Like I wrote in the subject, I don't think it's that important (for example
>>> to hold off 2.9.1), but still a bug. Not a very important bug even, but I
>>> can't get it out of my head that it's still a bug :)
>>>
>>> Can be fixed by making sure segmentSize() does not thrown any FNF exception
>>> for missing .prx? Does not sound too safe to me. Maybe we just remember this
>>> and tell people (like me) "even in silly tests, add a term to the document"?
>>>
>>> Shai.
>>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: java-dev-unsubscribe@lucene.apache.org
>> For additional commands, e-mail: java-dev-help@lucene.apache.org
>>
>>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: java-dev-unsubscribe@lucene.apache.org
> For additional commands, e-mail: java-dev-help@lucene.apache.org
>
>

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


Mime
View raw message