> Yeah, sorry, i think that WAS the point of your original mail, but it got lost in the conversation somehow... most of us take the architecture advice of web servers to heart.. be stateless to be scalable .. :)
I accept cookies if that's any help. :) > But anyways... It is, in general, not possible to stream the writing of an excel file, since the file structure contains upstream pointers to downstream data. Yeah, I kind of suspected that. > The changes in HEAD do not address this issue.. all that they do is optimise the memory usage of POI , ensuring that the same data is not stored twice etc. It should cause a 30-40% reduction in POI memory as well as time for large workbooks. On the subject of memory optimization--looking at the code that actually writes a workbook to a stream (HSSFWorkbook.write method), I noticed that it serializes the whole document into a byte array (using Record.serialize method), then copies the byte array to the OuputStream. It occurred to me that one could do away with the byte array and simply have the serialize method write directly to the OutputStream--seems like that could reduce the memory usage by quite a large factor. Does that make sense or am I missing something? > The solution for streaming like access currently proposed is to user the RandomAccessFile. Chris has proposed such a solution.. but not many people have actually worked with that solution. Doesn't sound like a RandomAccessFile would work for me--I want to stream directly across a network... Best regards, Robert Lowe http://RMLowe.com/ --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
