Closed mcr closed 3 years ago
https://github.com/dwbuiten/go-ffv1/blob/master/ffv1/rangecoder/range.go
also change reference from 3.8.1.1 to cite figure 9.
@dericed
Worth noting here that when I implemented go-ffv1, it was with an older revision of the spec, and it seems the figure numbers have since changed... I'll update the comments.
(Also worth noting that go-ffv1 is not an official reference decoder.)
Hi @dwbuiten, you could contextualize your comments to the version of the draft such as "draft-ietf-cellar-ffv1-17", although as the document is in "IESG Evaluation" status the RFC number may be soon.
Yes, I'll do that when I update them.
@dwbuiten, @michaelni, are you able to assist with the get_rac piece of Barry's review. The quote from the review is
— Section 3.8.1.2 —
"get_rac" returns a boolean, computed from the bytestream as
described in Section 3.8.1.1.
I see nothing in Section 3.8.1.1 that describes get_rac.
I understand that get_rac refers to figure 9; however a narrative could help. For instance does rac
mean RAnge Coded binary value
? If figure 9 is read as the i-th range coded binary value equals
, then what is the R_i
of the formula mean?
I understand that get_rac refers to figure 9; however a narrative could help. For instance does
rac
meanRAnge Coded binary value
?
IIRC get_rac() was named for consistancy with get_cabac(), it returns a single bit coded using the range coder. Not sure if the 3 letters can be matched to grammatically 100% fitting words
It looks like this issue can be closed (https://github.com/FFmpeg/FFV1/pull/230 and https://github.com/FFmpeg/FFV1/pull/241 are merged).
As per Barry Lieba review, the section 3.8.1.2 says:
but it is not defined, and likely relates to the decode process.