On Tue, Apr 17, 2001 at 08:34:07AM -0700, [EMAIL PROTECTED] wrote:
>...
> > Take a look at server/protocol.c, where this stuff came from. In between the
> > two sets of the timeout, you'll see it goes to the socket to read the
> > request line. Basically, it sets the timeout to <something>, reads a line,
> > then resets it to <something>.
> >
> > In the above code, it simply sets it a couple times and exits. Not very
> > effective :-)
>
> I'm fixing this now. I'll repost when I am done. This is part of the
> problem with splitting a large patch into a couple of small patches. Not
I hear ya. But I never saw this in the big patch, so having the smaller
patch is, therefore, quite a decent payback.
> that I am convinced it works correctly in the large patch, but I know I
> tackled this exact bug back when I first wrote the large patch. Expect
> the new patch later today.
I just realized one approach to fixing this.
The timeout surrounds read_request_line(). Well, that function is HTTP
specific. (not all protos read a "request line" as their first order of
business, and certainly not an HTTP request line)
Maybe you could find a way to move the timeout setting + read_request_line
to mod_http? As a bundle, this would work.
Cheers,
-g
--
Greg Stein, http://www.lyra.org/
