Re: Stats v2 proposal - for Thursday's meeting

On 2012-12-12 11:47, Harald Alvestrand wrote:
> On 12/12/2012 11:18 AM, Stefan Hakansson LK wrote:
>> On 12/11/2012 04:48 PM, Harald Alvestrand wrote:
>>> Enclosed is my stats v2 proposal, as it stands today.
>>> Things to note:
>>>
>>> - Using the dictionary *grammar* to describe objects, without making
>>> dictionary *objects* (since that's impossible to implement due to some
>>> WebKit limitations).
>>> - Using an unique "id" to point between different objects inside the
>>> stats DB.
>>> - Focusing on the objects that need defining, expecting to fill in the
>>> stats references later.
>>>
>>> I think this looks cleaner. It's obvious (to me, at least) where to add
>>> the stats that aren't included yet.
>>> It's only 6 pages, including the introduction and a code example.
>>
>> I think it would be cleaner if we did the object orientation thing we
>> did for DTMF for stats as well. So, instead of doing
>>
>> pc.getStats(selector, callbakcfunction);
>>
>> you'd do
>>
>> var statsReporter = pc.createStatsObject(selector);
>>
>> and then using the statsReporter object to obtain stats.
> Are you thinking of a new statsreporter object whenever one wants stats
> (so that the statsreporter provides access to a snapshot), or a
> statsreporter object that has only a getStats method?

I meant the latter.

>
> In the latter case, I don't see enough justificiation for the extra
> object, but this is a matter of taste.

Fair enough, the DTMF sender has twice the number of methods!

On the other hand, I can also see a development where we could add to 
PeerConnection the ability to set transport specific parameters per 
track in a similar way as DTMF sender. E.g.

var speakerVideoTransportHandler = pc.createControl(selector);

and then you could on that object have methods to set priority, BW, 
pause/resume sending, ... on that object.

Then it would IMO be cleaner to deal with stats in the same way.



>
>>
>>
>>
>>>
>>> Comments welcome - more on Thursday!
>>>
>>>              Harald
>>>
>>
>>
>
>

Received on Wednesday, 12 December 2012 14:42:27 UTC