On 1/23/06, Frank W. Zammetti <[EMAIL PROTECTED]> wrote: > Craig McClanahan wrote:
[snip] > > Calling flush() doesn't necessarily cause the *entire* response to be > > sent, because you might still be creating more output. But it *does* > > cause the HTTP headers, and any output content you've written to be > > transmitted. From the perspective of the servlet API, this response > > has now been "committed" so you cannot later decide to do a redirect > > or forward. > > > > In general, the only use case I can think of for calling flush() > > yourself is if you have something like a "Request In Progress - Please > > Wait" at the beginning of your page, and you want that to show to the > > user while a time-consuming process is performed to create the > > remainder of the content. > > Interesting... I would have thought just the opposite, but this makes > sense. I suppose one could imagine a problem with a filter that does > some post-processing occurring if you flush() yourself. > Yah, I suppose that's a potential use case, but I've never needed it. By the way, you also have control over how much output content is buffered before the container commits the response even if it's not done yet ... see methods like getBufferSize() and setBufferSize() on ServletResponse. You can also ask if the response has indeed been committed yet. > So, leave flushing to the next guy, contrary to what everyday etiquette > would seem to indicate ;) LOL > > > Struts doesn't but the servlet container will, once your processing > > has finished. > > This is a guarantee of the servlet API I would assume? It seems like it > logically would be, but I wasn't sure. Section SRV.5.5 of the servlet spec :-). > > > Craig > > Frank > Craig --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
