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

On 7/19/13 11:33 PM, Matthew Kaufman (SKYPE) wrote:
> I sure hope that Peter and folks with similar sentiment don’t
> **actually** think that the WG is requiring them to “wait until 2.0” to
> have these issues addressed.
>
> Among other things, if the first version can’t be implemented, it won’t
> be a standard… so it would be a shame to have not started on the correct
> API.
>
> I suppose if the WG chairs are really this inflexible to the membership
> of the WG, we can always start a new WG and get new chairs for it.

(This mail is not addressed to Matthew really, it is more for the entire 
list - and it is my personal reflection)

 From my perspective, we have been working towards one target for a long 
time (and as mentioned we had a poll last year that confirmed that), and 
it has not been that controversial. Sure, not everyone likes everything 
about the API - I'm pretty sure that no-one likes every aspect of it (I 
for one have some issues) - but that is what you often get out of a 
community effort.

There are also implementations of the to-be standard available. They are 
not behaving exactly the same - but that is to be expected with early 
implementations of something that is not yet specced up. And in addition 
there are apps using those implementations.

For the last month or so there has been a lot of traffic expressing 
dislike, and the need for a re-design. I think that the responsible 
thing to do in that situation is to understand more on why, and how 
things should be changed to be better. To throw everything out and start 
from scratch would not be fair to those who have implementations or to 
those using the implementations.

And, am I the only one fearing that starting over could take a long long 
time? A lot of people think the current API is really bad, but that does 
not mean that it will be easy to agree on what a good API should look like.

So the chairs have asked for use-cases that motivate changes. We've 
asked for API methods in addition the the current ones that there is 
need for. And we have invited to a discussion on v2 of the API, but 
proposed to start from use-cases. I don't think that is unreasonable, 
but also hear that use-cases is not the right starting point.

If you have ideas or proposals (being API issues, API proposals, 
procedures, use-cases, etc.), please submit them to the list (regardless 
of if they are for v1, v2 or vN) - I'm pretty sure that good ideas will 
be picked up. Perhaps someone has a really clever idea on a mod of the 
current API that would be seen as an improvement by most?

I fully agree to Matthew that currently there is not enough detail to 
build interoperable implementations in the combined (over W3C and IETF) 
set of documents, in particular SDP details are still lacking. That must 
of course be addressed.

Stefan

>
> Matthew Kaufman
>
> *From:*Peter Thatcher [mailto:pthatcher@google.com]
> *Sent:* Friday, July 19, 2013 12:29 PM
> *To:* cowwoc
> *Cc:* public-webrtc@w3.org
> *Subject:* Re: On babies and bathwater (was Re: [rtcweb] Summary of
> Application Developers' opinions of the current WebRTC API and SDP as a
> control surface)
>
> On Fri, Jul 19, 2013 at 12:15 PM, cowwoc <cowwoc@bbs.darktech.org
> <mailto:cowwoc@bbs.darktech.org>> wrote:
>
>     On 19/07/2013 3:06 PM, Adam Roach wrote:
>
>         On 7/19/13 13:47, Peter Thatcher wrote:
>
>             I think this is the real issue at hand: You value legacy
>             interop more than a usable API.
>
>
>         This. *This* is why I've told you that you're misunderstanding
>         everything. Don't take offense, just go back and read more
>         carefully. I never said anything that implied that legacy
>         interop is more valuable than a usable API. It's a nice strawman
>         for you to build up and tear down, but you are arguing with a
>         fictional character who is not me when you do so.
>
>         What I'm trying to point out is that these goals are not at odds
>         with each other. Your statement above implies that you have
>         taken it as given that we can't do both -- that there is a
>         tradeoff here to be made. If you take that as a fundamental
>         principle, then I can see how nothing I say makes any sense.
>
>         But they're not mutually exclusive goals. Keep that in mind, and
>         go back to re-read what I've written.
>
>          I think what Adam is trying to say is:
>     http://www.joelonsoftware.com/articles/fog0000000069.html
>
>          Adam, I would argue that we can improve the API incrementally
>     without throwing out all the lessons we've learned to date *but*
>     this means you have to be open to change.
>
>
> Ironically, the thing that seems to have gotten the whole SDP discussion
> reignited was that I proposed the "NoPlan JS API", which was a
> completely incremental, additive approach to improving the API.  So, I'm
> totally in favor of an incremental approach to improving the API.  But,
> when I proposed it, I got three big pieces of feedback:
>
> 1.  The anti-SDP crowd said, basically, "We don't like it because
> there's still too much SDP;  Remove more".
>
> 2.  The pro-SDP crowd said, basically, "We don't like it because this
> isn't SDP Offer/Answer;  Don't change so much".
>
> 3.  The WG leaders said, basically, "let's finish the current API before
> we change anything major".
>
> My hope is that 2.0 will be able to make everyone happy.
>
>     It's not okay to use this as a club to silence calls for change,
>     which frankly is what we've been hearing for a while: "Sit tight
>     while we finish 1.0... it's just around the corner. Let's not
>     discuss any changes until we get this out the door."
>
> FYI, Adam just said we're "nowhere near finished yet".
>
>     Gili
>

Received on Monday, 22 July 2013 08:17:59 UTC