I found some memory leaks in my test and it turns out that the come from ConjunctionScorer. Originally it freed its scorers in its destructor, but now this is commented out. Any idea why?
Jiri PS: When removing the comments memory leaks disappear. -----Original Message----- From: Itamar Syn-Hershko [mailto:ita...@code972.com] Sent: Monday, August 09, 2010 6:31 PM To: clucene-developers@lists.sourceforge.net Subject: Re: [CLucene-dev] cloning/modifying existing documents? Even Better - I may be wrong, but it worth a shot: Perhaps you can find the document(s) you need to update, and use that same document object with a call to IndexWriter::updateDocuement()? Even if it doesn't work right away, you could use void *updateDocument*(Term <../../../../org/apache/lucene/index/Term.html> term, Document <../../../../org/apache/lucene/document/Document.html> doc, Analyzer <../../../../org/apache/lucene/analysis/Analyzer.html> analyzer) and pass it a "dumb" analyzer, that uses a tokenizer that scans the term vector, and just approves all the tokens as they are (practically copying the term vector). Itamar. On 9/8/2010 7:19 PM, Itamar Syn-Hershko wrote: > On 9/8/2010 4:01 PM, John O'Brien wrote: > >> Hi, >> Apologies if this has already been covered in previous posts but >> I've not been able to find the answer in the archive so far. >> >> We have an application which indexes mail messages. We get the >> information for each message over IMAP, create the fields (e.g. subject, >> body, folder etc) and write the documents to the index. When a mail >> message is moved from one IMAP folder to another, our application gets >> notified of the move and we want to update the folder field in the >> existing document, so we create a new document, delete the existing one >> and write the new one. What I'm wondering is how other people use >> existing documents to create new ones - at the moment we get all the >> information over IMAP again which is obviously very inefficient but to >> make it more efficient we are now going to change it to retrieve all the >> fields and terms for the existing document and create the new document >> using them. Is there another (better/more efficient) way of doing this >> than retrieving the fields and terms for the existing document? >> >> > I guess you could use stored fields (Field::STORE_YES) to have this data > handy? it will make your index larger, but will prevent you from > retrieving the data again over IMAP or re-constructing it using the term > vectors (which might not work correctly for some analyzer > implementations). You can also use compressed fields if this is a lot of > data (watch out though, they have been deprecated in Java Lucene 2.9 or > so, and I assume we will have to update CLucene accordingly when time > comes). > > Itamar. > > ------------------------------------------------------------------------------ > This SF.net email is sponsored by > > Make an app they can't live without > Enter the BlackBerry Developer Challenge > http://p.sf.net/sfu/RIM-dev2dev > _______________________________________________ > CLucene-developers mailing list > CLucene-developers@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/clucene-developers > > > ------------------------------------------------------------------------------ This SF.net email is sponsored by Make an app they can't live without Enter the BlackBerry Developer Challenge http://p.sf.net/sfu/RIM-dev2dev _______________________________________________ CLucene-developers mailing list CLucene-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/clucene-developers ------------------------------------------------------------------------------ This SF.net email is sponsored by Make an app they can't live without Enter the BlackBerry Developer Challenge http://p.sf.net/sfu/RIM-dev2dev _______________________________________________ CLucene-developers mailing list CLucene-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/clucene-developers
