On Sun, Jan 27, 2002 at 08:54:31AM -0800, Brian Pane wrote: > That definitely would be less messy than an apr_brigade_destroy() > before each return statement. My only concern is whether recreating > the brigade would negatively affect performance. But if you can get > the same speed with that approach, I'm +1 on using it.
Well, consider that you're going to be calling ap_get_brigade in that one special case (folding) - i.e. you're going to be going through the filter chain again. So, I can't imagine that an extra brigade creation call should noticably hurt performance. And, the code will become much cleaner. I'll try to change this Monday night if no one beats me to it. -- justin
