Please provide patches relative to HM-dev.

Provided patches of minimal relevant changes. The reasoning to exclude "almost-the-same" naming from the reference s\w.

Btw, I don't know is it worth fixing this and if it is too late, but in JCTVC-Q1003 and JCTVC-Q1005 the same syntax element is called "...extension_flag" either "..._extension_present_flag".

Please check against R1013_v6 (latest version 2 document) for
"...extension_flag" and "...extension_present_flag".

Yes, there is one more flag is split from 8-bit chunk. However, my concern was that "pps_extension_flags" encountered in s\w were called instead of "pps_extension_6bits". For analysis among several hundred of syntax names the search for the match is painful, while fixing the s\w once is not of that.

Based on the patches I understand what you mean.

Usually I prefer to have HLS names aligned to the code, especially the trace strings. But in this case the current code looks much more flexible with the intention of the first eight extension bits. The multi-layer extension (as seen in the syntax in R1013_v6) can easily added in another "case" statement. We need to support only the RExt extension in HM, but SHM and the MV/3D software can easily extend the structure here (maybe even SCC with another extension).

I don't like the renaming of the array to pps_extension_Xbits, because it does not represent the standard text either. The only option that I would consider is to follow the syntax exactly and parse the already defined existence flags one by one and then the remainder. But with mentioned better flexibility, I would prefer to keep the code as is.

Then, my proposal boils down to move naming to R1013_v6 case, and udpate naming along text fixing; 4 times per year. Unlike PTL section, where 4 different syntax elements are named as the same, the extensions marking part is twice shorter and should not take too much efforts.

Could you please clarify the following (and related if something is):

JCTVC-R1013_v6 contains

vps_base_layer_internal_flag
vps_base_layer_available_flag

But the code has

READ_CODE( 2, uiCode, "vps_reserved_three_2bits" ); assert(uiCode == 3);

When it is supposed to change the code to match?

Fixed in r4252. A new ticket (#1365) has been created to highlight that some renaming of variables and functions is needed in order to be consistent with the text.