On Wed, 7 Mar 2001 09:03:14 -0800 (PST), [EMAIL PROTECTED] wrote:
>C-L is horrible. :-)
Yeah, this is getting really hairy....
>I have just commited something, which should stop the C-L filter from
>buffering unnecessarily. Could EVERYBODY please check my logic here. The
>commit message and the code have a block of text that says what I think
>should happen.
It's better (kinda) though now it sends a possibly incorrect content length
(and doesn't chunk) if there are no tags in the first 9k of the shtml. I
don't see how any response that goes through the include filter can have a
content length header unless the entire response is in a single brigade.
mod_include needs to kill the C-L as soon as it sends a partial response
down the stack.
>Brian, this should fix the underlying cause of the problem on OS/2.
The OS/2 specific problem is gone with my patch to apr_sendv(). Now it's
all non-mmap systems (I don't actually looked at why mmap gets around this
but others have said mmapped systems work better).
--
______________________________________________________________________________
| Brian Havard | "He is not the messiah! |
| [EMAIL PROTECTED] | He's a very naughty boy!" - Life of Brian |
------------------------------------------------------------------------------
