Re: Problems I found with ZOIE

[John Wang]

Hi Zhuobin:

 

    2) is fixed, see the bug for the link to the commit.

    I read over the code for 1), and I think what you are saying is indeed valid. 

    I opened up a ticket: https://linkedin.jira.com/browse/ZOIE-108 , and we can track it from there.

Thank you for your contribution!

-John

    

On Mon, May 28, 2012 at 11:39 AM, John Wang <john...@gmail.com> wrote:

Thanks Zhuobin!

I will get 2. fixed soon.

As for one, as long as the synchronized block are on the same monitor, it is ok. Synchronized blocks are reentrant.

-John

On Mon, May 28, 2012 at 2:29 AM, Zhuobin He <zhuob...@gmail.com> wrote:

Hello, I found two problem using the ZOIE 3.1.0 that want to confirm with you:

1. In RealtimeIndexDataLoader, the consume() method add the doc into ram index within a synchronized block. And the processBatch() method which responsible for merging ram index to fs index is also in synchronized. I think these two method will block each other and thus zoie will lose realtime feature while merging ram index to fs.
2. ZoieSegmentTermDocs doesn't reset the _nextDelDoc cache after seek. This may cause bug if the ZoieSegmentTermDocs are reuse in one query for multi term, such as term prefix query, in which case it may treat some deleted doc as nondeleted.I have report a bug on jira https://linkedin.jira.com/browse/ZOIE-106