Re: Lucene Input/Output error

2008-10-25 Thread JulieSoko
I have been doing some reading and think something that I am doing icould be a problem. For each search of an index, I first check to see if the IndexReader is current and if not reopen it... I think I should not reopen the IndexReader until I make sure that there are no other queries running u

Re: Lucene Input/Output error

2008-10-25 Thread JulieSoko
I'm using Linux... running out of file descriptors... I didn't think of that! I will check on Monday... I forgot to mention that I am using the Lucene 2.3.1 version. Not sure if that makes any difference. Could there be an issue with the indexes being merged and searched at the same time? Mich

Memory problem dealing with indexsearcher and topdocs

2008-10-25 Thread Marc Sturlese
Hey there, I am having some memory trouble with my Lucene app. I need to get the info and delete about 1000 docs every time I execute the app. I get the IDs of the documents to delete from a database and for all single ID I get the data from the indexed doc using an index searcher and topdocs (sea

Re: Combining keyword queries with database-style queries

2008-10-25 Thread Niels Ott
Erick Erickson schrieb: Is this an inadvertent re-post or is there still something you're wondering about? I originally posted with an address not subscribed to the list, causing the list server to silently swallow the message. Now the message apparently made it here with some delay, for what

Re: Lucene Input/Output error

2008-10-25 Thread Michael McCandless
What OS/filesystem are you using? The code looks fine to me. 2 or more searches running at the same time on the same index should be harmless; this happens in Lucene all the time in normal applications. Are you sure you're re-using already opened Searchers, and not accidentally opening

Re: Possible payload bug in lucene

2008-10-25 Thread Michael McCandless
The problem is you are not first calling tp.next(). If you fail to call tp.next() then tp.nextPosition()'s behavior is undefined. In a multi-segment index, you'll hit the NPE you're seeing. In a single-segment (fully optimized) index, you don't hit that exception but I'm not sure if the

Re: Multiple values in field

2008-10-25 Thread agatone
Yeah but if it happens that for a certain document field1 has only one value and in that case I can't know it is intended as a multiple. in that case adds.size() would be == 1 and it would look as normal field. What i need is to mark that field is intended to be multiple no matter if it contains