Andrew C. Oliver wrote: > 4. Once we get the object creations killed we should work on a memory/file > map system. For this to work we have to touch the lowest level (the POIFS > code). This will require someone to become a very high-caliber low-level > developer. I can't take this one on myself. We need someone who wants to > step up to like the most challenging thing. Unfortunately, it will be hard > to get funding for something here (just laying my cards out) as I can't > explain it to clients very well (they touch HSSF, not POIFS). This won't be > usable in all situations and you're going to be disappointed with the > performance (but you can think I'm wrong for now) but it will be more > efficient with memory... Its also WAY more work than you think. Once POIFS > has memory/file mapping we can do it on up the line for HSSF/HWPF/etc.
In the pre-POI era our company developed a Java OLE library similar to POIFS, as part of a PowerPoint generation library. We use POI too and it would be better for us if everything worked against a common codebase, so we're interested in contributing if it makes sense. The library uses a "Seekable" interface for which there are byte[], InputStream, and RandomAccessFile (+mmap) implementations. It is similar to some of the imageio API. This chunk is actually pretty generic, and might even be worth putting into Commons IO. Let me know if you have any questions, and what you think the process for integrating this should look like. Thanks, Chris Nokleberg Tonic Systems, Inc. --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
