In the section 8.3.3 of the latest spec (JVET-R2001-vA or JVET-S0152-v5), there is a paragraph that "STRPs and ILRPs are identified by their nuh_layer_id and PicOrderCntVal values. LTRPs are identified by their nuh_layer_id values and the Log2( MaxLtPicOrderCntLsb ) LSBs of their PicOrderCntVal values." There are two issues here.

1. "MaxLtPicOrderCntLsb" is not defined in the spec, it should be changed to "MaxPicOrderCntLsb"
2. In current RPL design, LTRP is not always identified by LSB, it may also be identified by full POC value.

So it is suggested to change to "LTRPs are identified by their nuh_layer_id values, and the Log2( MaxPicOrderCntLsb ) LSBs of their PicOrderCntVal values or PicOrderCntVal values"

In second paragraph of section 7.4.5,"num_ref_pic_lists_in_sps[i]" should be changed to "sps_num_ref_pic_lists[i]".

1.) Enabling JVET-P0551 disables the macro JVET_O0625_ALF_PADDING. This has the side-effect of disabling the JVET_P0158_ALIGN_ALF_VB changes.

2.) I believe the JVET-P0551 isCrossedByVirtualBoundaries( ) slice boundary detection may be missing logic to detect and pad top-left or bottom-right slice edges. For example, when the top and left CTU are both within the current slice, but the top-left CTU is not. These cases can happen with raster-scan slices.