failing to loose route sequential requests that are proto wss

[etamme]

On master commit 968a4e0

If I place a call from a WSS client to a UDP client, then the UDP client hangs up, opensips will show the following log indicating that it cant identify the protocol.

Jan  5 12:35:23 voipphreaker /usr/local/opensips/sbin/opensips[9885]: DBG:rr:after_loose: Topmost route URI: 'sip:159.203.234.107;r2=on;lr;ftag=s6992f0ms9;did=e6a.94fb3f15' is me
Jan  5 12:35:23 voipphreaker /usr/local/opensips/sbin/opensips[9885]: WARNING:core:grep_sock_info: unknown proto 9
Jan  5 12:35:23 voipphreaker /usr/local/opensips/sbin/opensips[9885]: WARNING:rr:after_loose: no socket found to match 2nd RR [9][159.203.234.107:443]

I dug around in sock_info.c and net/trans.c to find how the protos list was populated, but I'm not able to immediately see whats wrong. get_sock_info_list appears to be returning 0, but it looks like the wss proto is in the list as proto 9 as the log indicates, so I'm not clear on whats going on.

[etamme]

This has been tracked down to parse_uri.c not parsing wss - looking into how to fix it now

[etamme]

im close on a patch that fixes this, but im having trouble because WS and WSS are the same letters up till the last S. Im not sure how to avoid the transport_fin call for WS as i basically need to look ahead.

here is my current diff https://gist.github.com/etamme/953cb286cb73b6a62216
If you uncomment

//transport_fin(VW_S_FIN, PROTO_WS);

It will fail to compile, but i tested it as is with WSS and got clean loose routing on a sequential request towards a WSS client.

[bogdan-iancu]

Thanks to @jarrodb , this is fixed via #744