JVET_C0024_QTBT switched off with corresponded configurations cause encoder's crash. It works normally if JVET_C0055_INTRA_MPM is also switched off.

In function 'TEncSearch::xEncIntraHeader', the function call inside macro COM16_C1046_PDPC_INTRA should be after the closing bracket (}), instead of before it.

With the bug, the function `m_pcEntropyCoder->encodePDPCIdx' is called only for non-intra slices, when the correct is that it should be called for all slices.

The bug's effect in coding performance is minor. A file with the corrected version is attached.

In the JEM encoder there seems to be a problem in that some motion vectors that are evaluated becomes larger than what is possible to represent with "short". This causes an incorrect (overflown) motion vector to be evaluated. It does not seem to create any serious problem in the main branch but in the EE5 branch it currently causes the encoder to crash on some sequences. To me it seems it would be better to consistently use "int" instead of "short" for the motion vectors so that no overflown motion vectors are evaluated. Making that change might be a better solution than making a local bug fix in the EE5 branch to avoid the crash. But since it is probably a quite substantial modification to the code (changing all motion vector occurrences from "short" to "int") I would like to hear others' opinions and suggestions.