2010/8/11 Kostka Bořivoj <kos...@tovek.cz>: >> For the patch of the RAMFile desctructor I'm in doubt. Looking at the >> correspondong >> JLucene classes subtracting the file size is done in the deleteFile method of >> RAMDirectory (see diff). >> >> By the way, I needed some inline keywords removed in _RAMDirectory.h. >> Otherwise, I couldn't compile it with gcc on FreeBSD. > > I agree, it is better to do this in deleteFile than in RAMFile destructor. > The only thing I'm not sure about is the mutex used for sync - files_mutex > is used in deleteFile. > Shouldn't it be RAMDirectory's THIS_LOCK?
Maybe. But every other method manipulating the files map is using the files_mutex also. >> > 2) MockRAMDirectory::close always throws an exception because >> > segments.gen is in list of open files, no idea if this is a bug in >> > RAMDirectory or MockRAMDirectory >> >> Why do you use MockRAMDirectory here? I ported this class from JLucene, >> because I >> needed once a directory implementation that intentionally throws exceptions. >> And >> because the close-methods of RAMDirectory and MockRAMDirectory are identical >> to >> the JLucene methods, I'd say that's neither a bug of RAMDirectory nor >> MockRAMDirectory. > > I use MockRAMDirectory just because it is used in original java test I > ported. It is used > in many tests in test/index folder so I wanted to test MockRAMDirectory > itself before > using it in other tests. You are right. Something is odd there. I will take a close look tomorrow. Veit ------------------------------------------------------------------------------ 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
