On Wed, Jun 1, 2022 at 12:39 PM Ruediger Pluem <rpl...@apache.org> wrote:
>
> On 6/1/22 11:56 AM, yla...@apache.org wrote:
> >
> > + /* Compute Host header */
> > if (dconf->preserve_host == 0) {
> > if (ap_strchr_c(uri->hostname, ':')) { /* if literal IPv6 address
> > */
> > if (uri->port_str && uri->port != DEFAULT_HTTP_PORT) {
> > @@ -3994,10 +3992,11 @@ PROXY_DECLARE(int) ap_proxy_create_hdrbr
> > host = uri->hostname;
> > }
> > }
> > + apr_table_setn(r->headers_in, "Host", host);
> > }
> > else {
> > - /* don't want to use r->hostname, as the incoming header might
> > have a
> > - * port attached
> > + /* don't want to use r->hostname as the incoming header might have
> > a
> > + * port attached, let's use the original header.
> > */
> > host = saved_host;
> > if (!host) {
> > @@ -4007,10 +4006,9 @@ PROXY_DECLARE(int) ap_proxy_create_hdrbr
> > "on incoming request and preserve host set "
> > "forcing hostname to be %s for uri %s",
> > host, r->uri);
> > + apr_table_setn(r->headers_in, "Host", host);
> > }
> > }
>
> Nitpick: Can't we do the apr_table_setn(r->headers_in, "Host", host); here
> instead of in each if/else branch?
This is a small optimization where if we reuse the existing Host
header (saved_host) we don't need to set it again.
But if it harms readability I can certainly change it as you say.
Regards;
Yann.
