Re: [Gnash-dev] NetConnection And NetStream

[Rob Savoye]

On 02/27/13 09:15, George Thomas wrote:

> Will have a look at understanding the existing support and will try to get
> the code organised so that everything works.

  You can bounce RTMP questions off me, I did the reverse engineering of
this protocol, and wrote our first implementation, which is the one we
should be using.

> I was thinking of using red5 as server and getting tests running with it
> and adding cases to regression framework during the course. Also found
> cygnal to contain code for RTMP but did not explore completely.

 Cygnal does support RTMP remoting, as well as Red5. They even can share
client side test cases. Red5 will work as well.

> Is it the best way to go about things ?

  Setup a test environment, there are also some docs on our wiki. Other
info can be found on youtube, look for my talk at FOSDEM on "Reverse
Engineering of Proprietary Protocols".

> Also in that case shold we expect ValueType.tNull (as in
> classes.all/net/NetStream_as.hx) to be return type or should it be same as
> that of the prototype ?

  So I had the same question yesterday when fixing this test case. Haxe
sets these prototypes to TNull by default, then seems to use the correct
TFloat, TBool, or TInt when they have real values. Ideally we need to
test the test cases with the proprietary player. I was just trying to
get rid of the new FAILs.

        - rob -