On 8 Mar 2001, Jeff Trawick wrote:
> I don't see the connection unless there is some bug in the ring macros
> which adds arbitrary buckets to the ring/brigade.
>
> I don't see why an eos bucket would even be in the brigade here. I
> would have thought that the HTTP input filter would do a read(), grab
> the rest of the data from the client, and before thinking it should do
> another read() (to hit EOF) it would return a brigade to the caller.
I'm not saying that there aren't possibly more than one bug in ap_getline
(in fact, I've suspected that corefiles 4 and 9 were caused by two
slightly different problems). I'm just pointing out that they all go
together, at least in as much as it's the same assertion that's failing.
--Cliff
--------------------------------------------------------------
Cliff Woolley
[EMAIL PROTECTED]
Charlottesville, VA
