Re: [rtcweb] On babies and bathwater (was Re: Summary of Application Developers' opinions of the current WebRTC API and SDP as a control surface)

[Martin Thomson <[e-mail]>]

On 19 July 2013 10:39, Peter Thatcher <[e-mail]> wrote:
> So, are you trying to say "Look how hard this is to do with SDP.  We're not
> even done yet.  It will be even harder without SDP"?

In the Atlanta meeting (Nov 2012) I remember waiting for those damned
elevators with Justin.  He said something like: "So, if we'd chosen
comment 22 [CU-RTC-Web] do you think we'd be done by now?"  I was
quick to answer, "Of course not."  After all, we'd only made that
proposal 3-4 months earlier.  I know that nothing gets done in less
than a year, and this is not a small undertaking.

Since then, I've gained a more nuanced view.  We've set an impossibly
high bar for this specification, including all sorts of crazy
features: FEC, simulcast, layered codecs, congestion control,
undeployed codecs, multiplexing, and not to mention a new data
channel.

In reality, SDP never negotiated all that crap before.  Worse, despite
the existence of RFCs for most of these features, it turns out that
most implementations were proprietary.  We couldn't even agree on what
an m-line represents.

So, if asked the same question today, I'd probably have to say: "We'd
at least have had basic scenarios shipped.  We might not have sorted
out the hard cases like layered codecs, but we do have basic
functionality."

What we have this the complete antithesis of any project I've been
involved in over the past 10 years.  It's gold-plating the pink
squirrel.

If we want to ship this thing, then we should be managing scope, not
protecting it.