I've been hacking furiously on RTMP* support in Gnash, so I can use it
for Cygnal testing, and other things. Currently this has just involved
refactoring the remoting support in Gnash to use the RTMP support
libraries I've been building. These also support RTMPT (AMF over
HTTP),
and work both client side and server side of these network protocols.
To make sure this worked, I enhanced the red5test.as test case to
support testing with dozens more data types, including nested ones.
As I
had been using the regular red5 client side echo tests for Cygnal, I
knew that hex data very well, so I ported most of the echo_test.swf
tests to Ming.
There are now RTMPT and RTMP versions of the same tests, and I added
support for automatically testing Gnash using the network. I added
support to the test driver to use a list of possible servers, starting
with localhost, then gnashdev.org, (I installed Red5 there), etc...
These tests are currently designed to be used with Red5 for now,
although they work with Cygnal too. I'm currently changing Cygnal to
run
the echo response as a separate process, since this is how it'll
need to
work in the future to support server side ActionScript.
It's been interesting working on the client side of RTMP, I found a
few
subtle tweaks to make to Cygnal for better compatibility. So far it's
been relatively stable, handling bad packets, corrupted data, network
latency, etc... well.
What doesn't work yet is RTMP support for NetStream, that's the next
big
task. The current plan is to finished the final debugging of the rtmp
branch (more potlatch testing, rtmpget), migrate all this to trunk,
and
then launch off on that. Hopefully later in April I'll have Cygnal
successfully streaming video to Gnash via RTMP. :-)
- rob -
_______________________________________________
Gnash-dev mailing list
address@hidden
http://lists.gnu.org/mailman/listinfo/gnash-dev
PGP.sig