Coming to think of this, I suppose we should not pose any upper limit in RAMDirectory. Environments with >4GB of RAM could use this feature.
I couldn't find the Exception I referred to in my previous email, and since you say you have it working now it probably ain't there anymore. So I think your change makes sense. Can you test to see if using int64_t this way on a 32bit platform works as well? and what happens if you don't use this typedef on both platforms (use just int or long instead of just int64_t)? Itamar. On 11/7/2010 6:07 AM, Liu bbskill wrote: > I used he RAMDirectory in 64bit machines. > > And in RAMDirectory, I modified the RAMIndexInput/RAMIndexOutput's > pointer variable to be int64_t, and also modified a little codes of > RAMIndexInput::readInternal and RAMIndexOutput::flushBuffe to coincide > with it. > And I succeeded in loading and searching the 5.8G index. It seems work > fine for me. > > And I am curious about whether this modification is just correct.. Am > I missing something? > > 2010/7/11 Itamar Syn-Hershko <ita...@code972.com > <mailto:ita...@code972.com>> > > On 9/7/2010 11:02 PM, Veit Jahns wrote: > > That's an internal limit of Java Lucene (see e.g. [1]) as well as > > CLucene. That's all I know about this, but Michael and Itamar > > discussed about a similar issue regarding FSDirectory some time ago > > [2]. May be this helps you with this issue. > > > From what I can tell, you are hitting the internal limit of 2GB > hardcoded into RAMDirectory and not that other issue Veit has pointed > to. You shouldn't be using RAMDirectory for 2GB files really. I don't > think even (N)RT searchers ever do that. > > What Michael and I discussed was related to an issue with FSDirectory, > where >2GB files weren't read/written to correctly in 32bit systems. > > I can't remember if anyone ever got this fixed. If someone could > have a > look at it again (Michael has provided an initial patch), it would > be a > great help. > > Itamar. > > > ------------------------------------------------------------------------------ > This SF.net email is sponsored by Sprint > What will you do first with EVO, the first 4G phone? > Visit sprint.com/first <http://sprint.com/first> -- > http://p.sf.net/sfu/sprint-com-first > _______________________________________________ > CLucene-developers mailing list > CLucene-developers@lists.sourceforge.net > <mailto:CLucene-developers@lists.sourceforge.net> > https://lists.sourceforge.net/lists/listinfo/clucene-developers > > > > > -- > JinBiao Liu > > > ------------------------------------------------------------------------------ > This SF.net email is sponsored by Sprint > What will you do first with EVO, the first 4G phone? > Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first > > > _______________________________________________ > CLucene-developers mailing list > CLucene-developers@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/clucene-developers > ------------------------------------------------------------------------------ This SF.net email is sponsored by Sprint What will you do first with EVO, the first 4G phone? Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first _______________________________________________ CLucene-developers mailing list CLucene-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/clucene-developers
