The existence and naming of ARR_MAX_HEADER_SIZE is somewhat dubious,
as it is:
Thanks you for the feedback. I cleaned up the patch.
* Used in exactly one place (not necessarily a reason why it should
not be reified into a stand-alone definition, though, but
something to consider)
Moved it to one definition
* The array header refers to the NULL bitmap as well, but the
interpretation used by the patch does not.
I renamed the macros to have NONULL in the name (hopefully it doesn't make them too long).
I also added a comment. Not quite sure if it's the appropriate format, but I didn't feel it warranted 3 lines.
Thanks,
Mike Lewis