NDT-DEV email list created

[Will Hawkins <>]

Hey Aaron!

Full disclosure: I haven't read your entire note yet. However, I wanted
to respond quickly and let you know that at OTI we are in the process of
developing a browser plugin to run NDT clients. We are leaning to
targeting Firefox first and their SDK promotes the use of Javascript for
developing extensions. The possibility is there for us to reuse that
code extensively and swap out websockets for the extension-only i/o
functions.

We are in the process of doing some performance testing to determine
whether the throughout of the SDK network i/o is fast enough to support
NDT. Once we are done with that, we will send out the results. Then, we
can compare it with the throughput of websockets.

Does that sound reasonable?

:-)

Will

On 04/11/2014 03:49 PM, Aaron Brown wrote:
> Hey Folks,
> 
> Someone asked about a javascript NDT implementation, and I was wondering 
> what folks thought about the feasibility of this. Based on my rather 
> cursory understanding of websockets, it looks like essentially a simple 
> socket that requires an initial handshake and that data to be formatted in 
> a specific manner. If we have the NDT server look for that initial 
> handshake in the initial connection (and in test connections), respond to 
> the handshake, and then pack/unpack the normal protocol back-and-forth into 
> websocket frames, theoretically, the entire thing could be handle with 
> essentially no major protocol changes. It wouldn’t be pretty by any 
> stretch, but it theoretically could(?) work.
> 
> Can anyone with more websockets knowledge throw a wet blanket on this idea, 
> or do other folks otherwise think it’s a dumb idea?
> 
> Cheers,
> Aaron
>