Re: [rtcweb] Opsdir telechat review of draft-ietf-rtcweb-fec-08

[Justin Uberti <[e-mail]>]

On Tue, Feb 12, 2019 at 11:57 PM Jürgen Schönwälder <
[e-mail]> wrote:

> Reviewer: Jürgen Schönwälder
> Review result: Has Nits
>
> I have reviewed this document and I did not find any issues that are
> affect the operations of networks (except that FEC requires more
> bandwidth and that it may not help much or even make things worse if
> bandwidth is the cause of packet loss, which is explained in the
> document).
>
> While reading the document, I wrote down the following notes that the
> editor may take into account when revising the document:
>
> - in 3.1, expland SSRC on first usage and say that this is about
>   sending streams over RTP somewhere early on (I assume this is
>   implied by using the term WebRTC but for outsiders like me it may
>   help to be more specific).
>

Acknowledged.

>
> - To what extend is this document WebRTC specific? Do the requirements
>   also apply if I use RTP without a WebRTC context? If so, should the
>   title rather say "RTP Forward Error Correction Requirements"? Well,
>   section 6 may be WebRTC specific but that section just says that
>   nothing is being recommended, so the recommendations are really all
>   about FEC usage over RTP as far as I can tell (as an outsider).
>

This document indicates what mechanisms WebRTC implementations should
support, so while the recommendations could also be applied to non-WebRTC
endpoints, that is not the focus of the document.


> - stylistic: I am not a big fan of using citations like '[RFC2198]' as
>   ordinary words or nouns, it makes text difficult to follow unless
>   you know the RFC numbers by heart and your brain translates them
>   back to something meaningful on the fly. This makes texts more
>   difficult to read for outsiders. Example:
>
>      This mechanism is similar to the
>      [RFC2198] mechanism described above.
>
>   I prefer this:
>
>      This mechanism is similar to the
>      redundant encoding mechanism described above.
>
>   There are couple of such usages of [RFCXXXX] in the document
>

OK.

- add reference for PCMU
>

OK. Reminder to self, RFC 5391.