[vlc-devel] Re: vlc: svn commit r17986 (md)
rem at videolan.org
Fri Nov 24 12:29:52 CET 2006
Le vendredi 24 novembre 2006 12:11, Marian Durkovic a écrit :
> > There is no regression that I can see. The last nightly branch
> > works fine on XP. And it is fixed finally on Linux.
> No it doesn't ! If you like to have this ugly hack for Linux, at
> least spend a few minutes in order NOT to break properly working
> RFC-compliant systems - e.g. by including your "fixes" into
> #ifdef UTTERLY_BROKEN_IGMPv3_IMPLEMENTATION
> It would still be much more apropriate to fix the Linux IGMPv3 code,
> since otherwise such ugly hacks will be necessary for every multicast
> application which is definitely NOT the way to go.
From https://trac.videolan.org/vlc/ticket/826, you reported, about the
specification of a server name for ASM streams:
"it will always trigger IGMPv3 source-specific joins if IGMPv3 is
"And regarding IGMPv3 - you apparently haven't debugged multiple IGMPv3
bugs in multicast routers from biggest vendors. I did. One as an
example - stream working fine with IGMPv2 interrupts every 5 minutes
So we made sure VLC would not try SSM joins for ASM streams. Now that I
remove SSM joins for ASM streams you complain that it should be brought
back. How am I supposed to tell which of your conflicting versions is
correct? If you decided that afterall IGMPv3 is good for ASM streams,
no problem, revert the ~18000 commits... but don't complain again if
VLC reintroduces source-filters for ASM streams later on.
As for the connect() issue with multiple-but-not-any sources, I agree
completely. Connect() cannot and must not be used. But VLC current
network plugins architecture does not support this anyway, neither does
the source-filter parser or the UDP access, so at least in that sense,
there is no regression since it has never been working.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Size: 189 bytes
Desc: not available
More information about the vlc-devel