> Besides, it is probably an HTTP violation to forward trailing garbage
> anyway, unless we switch to some kind of "tunneling mode" and become a
> TCP proxy, but then you get problems with message smuggling and such.
>
> Do you think this change should go in?

Trunk and 3.1, I think so, 3.0 I'm uncertain.
If the time allows, maybe protecting it with a fast-acl-backed
selector (defaulting to 'apply to all') would be the most flexible
solution, allowing (especially rproxy) administrators to fine-tune the
behaviour.


-- 
    /kinkie

Reply via email to