VP8 Data Format and Decoding Guide
RFC 6386

It is always a shame when a document shows up (even as an independendent publication request) and fails to pass idnits.

---

Thank you for the inclusion of the clear patent-related text in the document.

Informational documentation is good, especially when it is accurate and complete. I am unable to judge the accuracy of this document (however, I did notice some small errors, such as "/*!\file decoder_impl.h" instead of "/*!\file vpx_codec_internal.h" in Section 20.19). As Wes Eddy noted regarding completeness, some files seem to be missing. Although I'm not quite sure which files are supposed to be included, according to my count the following files might be missing:

coef_update_probs.c
dboolhuff.c
dboolhuff.h
decodeframe.c (in one place called decodframe.c)
decodemv.c
dmode.c
filter_c.c
onyxd_if.c
quant_common.c
reconinter.c
reconintra.c
tree_reader.h
vpx_decoder_compat.h
yv12config.h

I'm surprised there are no security considerations in all that text.
Given that codec implementation flaws have been fairly common, it would
be good to give guidance on this, e.g. perhaps a small amount of
generic guidance (buffer overruns etc.) but also some detail if there
are specific things in this codec that implementers are likely to get
wrong.

I have only reviewed this document for Routing implications and see no issues.

The document seems to be incomplete.  I'm unsure where to find the yv12config.h referenced in section 2 or the vpx_config.h referenced later -- are these supposed to be included in the document, or do they live somewhere else that the reader is supposed to know how to find?  You should make sure this is as self-contained as possible, as references to files that may  not be available N years from now doesn't seem like a good idea.  Note, there may be other files like this missing; I only spot-checked a number of them as I was walking through it.  Otherwise, I thought this was nicely written.