Editorial suggestion for picture_header_in_slice_header_flag single slice constraint

[Vadim]

In picture_header_in_slice_header_flag semantics, there is only one constraint "It is a requirement of bitstream conformance that the value of picture_header_in_slice_header_flag shall be the same in all coded slices in a CLVS."

It gives a wrong impression that it can be multiple slices in a picture where this flag value is the same.

However, in the section "7.4.2.4.4 Order of NAL units and coded pictures and their association to PUs" there is additional constrain that "When a picture consists of more than one VCL NAL unit, a PH NAL unit shall be present in the PU."

Suggestion is to add explicit constraint to picture_header_in_slice_header_flag semantics that when picture_header_in_slice_header_flag is equal to 1 it shall be one VCL NAL unit in a picture.

[yk]

Note that JVET-Q2001-vE also has the following constraint (in the semantics of picture_header_in_slice_header_flag semantics): When picture_header_in_slice_header_flag is equal to 1 for a coded slice, it is a requirement of bitstream conformance that no VCL NAL unit with nal_unit_type equal to PH_NUT shall be present in the CLVS.

That and the two constraints you listed above together tell that when picture_header_in_slice_header_flag is equal to 1 there can only be one VCL NAL unit in each picture in the CLVS.

Therefore, I think the issue herein is not valid. Unless concluded otherwise, I will close it (but after some more discussions, if any).