Multicast TTL confusion

There seems to be significant confusion about how to properly set TTL for multicast streams. There are 3 parameters:

--ttl
--sout-udp-ttl
--sout-rtp-ttl

If only --sout-xxx-ttl parameter is set, stream packets are flowing with specified TTL, but SAP packets with TTL=1. If --sout-rtp-ttl is not specified, the SAP announcement contains 0 as TTL field in c=line.

I'd tend to removing the --sout-xxx-ttl parameters completely and using --ttl everywhere.

A bit late for 0.8.4 . Needs some testing + string changes

changed milestone to %0.8.5-final

Update: after [12964] the c= line of SAP announcements from #rtp sout matches real TTL of RTP packets.

--sout-rtp-ttl (default 0) takes precedence over --ttl to be able to set different TTL for a specific stream.

TODO: All SAP announcements are sent with TTL equal to the value of --ttl when the SAP handler was started. As it lives until VLC exits, subsequent changes in --ttl are not propagated to the SAP handler anymore. Values in --sout-xxx-ttl are not propagated to TTL of SAP packets.

Sap should be 255 according to RFC

assigned to @Courmisch

(In [14207]) - Use OS default TTL when not overriden

• Clean up (refs #404 (closed))

Good point zorglub - so let's fix the TTL for all SAPs to 255.

Regarding [14207] - it seems that the default setting for --ttl in libvlc.h is 1 so VLC will by default use TTL=1. Shouldn't we set the default to zero to use OS default?

(In [14223]) Good point Md - use 0 (ie. not set) as default TTL rather than 1 (refs #404 (closed))

(In [14224]) SAP hop limit should always be 255 as per its specification - refs #404 (closed)

I could not find any mention of --sout-rtp-ttl anywhere... are you sure about that one ?

(In [14227]) - Fix typo in previous commit

• Enforce a default TTL value (one) in SDP if there is none - refs #404 (closed) (it should at least be fine for IPv4 usage)

(In [14228]) Handle TTL in SDP properly (IPv4 at least) for raw UDP (closes #404 (closed))

added Status::fixed label

closed

mentioned in issue #527 (closed)