Re: [AVT] RTP: explanation of "sampling instant"

[Stephen Casner <casner@acm.org>]

Dave,

> >>              Applications transmitting stored data rather than data
> >>              sampled in real time typically use a virtual clock to
> >>              determine when the next frame or other unit of each medium
> >>              in the stored data should be presented.
> >
> >I think "determine" is the wrong word here. Its the receiver that
> >determines when to present it. The sender indicates when it should
> >be presented. So, I would propose:
> >
> >Applications transmitting stored data rather than data sampled in
> >real time typically use a virtual clock to indicate when the next
> >frame (or other unit of each medium) in the stored data should be
> >presented.
>
> This has always been my problem about how this is presented;  it is
> an unusual protocol description which describes what happened at the
> source.  Usually protocols describe characteristics of the stream or
> what the receiver ought to do.  In particular, there is no 'sampling
> instant' for streams sent from disk under rtsp control, for example.
> However, there is a reference common clock which can be used to
> relate the RTP timestamps in the various streams.

I don't think it is a description of what happened at the source.  We
are choosing what particular reference point in the life of the data
represented by the RTP timestamp in the packet corresponds to the
associated reference clock timestamp.  Choosing a common reference
point in the life of the data allows data from separate sources to be
related.  Specifying some amount of time to be added to the sampling
instant in order to hypothesize a presentation instant seems
artificial to me, and adds an extra complication.

I've proposed some additional text to try to explain the motivation
for the choice of reference point, and to explain how "sampling
instant" should be interpreted for the case of streams sent from
disk.  Do you think the words I added made the situation better or
worse?  Do you have other words to suggest?

I already replied to the message you were replying to; I explained why
I used the phrase "determine when the next frame or other unit of each
medium in the stored data should be presented".  How would you
describe the process that the sending is doing when it uses a
reference clock to determine when it is time for the next piece of
data to be transmitted?

                                                        -- Steve

_______________________________________________
Audio/Video Transport Working Group
avt@ietf.org
https://www1.ietf.org/mailman/listinfo/avt