On Thu, Sep 19, 2013 at 5:32 PM, Plüm, Rüdiger, Vodafone Group <
[email protected]> wrote:
> What did you change from the original patch?
>
I changed :
- conn->keepalive = AP_CONN_CLOSE
,
according to your comment
,
-
APR_ENOTIMPL => APR_EGENERAL
in ap_http_filter
so that ap_map_http_request_error
returns 400 and not 501,
- APLOG_WARNING => APLOG_INFO when "using read-until-close".
- (strcasecmp(tenc, "chunked") != 0 && !ap_find_last_token(f->r->pool,
tenc, "chunked")) => !(strcasecmp(tenc, "chunked") == 0 ||
ap_find_last_token(f->r->pool, tenc, "chunked")); this one shouldn"t harm
and is more what the comment (draft) says,
- if (...) { fall through } else if (f->r->proxyreq != PROXYREQ_RESPONSE) {
return EGENERAL } else { fall through } => if (...) { fall through } else
if (f->r->proxyreq == PROXYREQ_RESPONSE) { fall through } else { return
EGENERAL }; shouldn't harm either, just to end with the failure condition.
I don't see why Jim's tests failed, even with the first patch, since the
case does not seem to be tested, so APR_ENOTIMPL => APR_EGENERAL is not hit
(and it's the only change regarding the response status)...
Maybe Jim did have some (other) local changes ?
Regards,
Yann.
