confirmed bug in db_init_chunk / db_chunk

[paul mineiro]

chris suggested that there might be a bug here but since he was using
a different version of mnesia we weren't sure if it was the patches or
a version mismatch.

i can confirm that i saw the remote table loader crash on the sender
side due to a tcerl select continuation being passed to ets by mistake
using mnesia-4.3.5 . we think we have the fix to
mnesia_lib:db_init_chunk/3 and mnesia_lib:db_chunk/2 but i want to
test it first. if it works then i'll release ext2.

-- p

[paul mineiro]

Hey I got this working on my setup, released as ext2.

The "attempt to insert the term 'copier_done'" into the external table
goes away with this fix, since it only happens when there is an error
during table transmission, but I still think there is a lingering
problem, because the spec for an init fun is very clear that it must
either return a list of objects and continuation { Objects, Cont }, or
else end_of_input; { copier_done, Node } is neither of these.

Oh well it's an error case, maybe it's just broken in mnesia and
nobody notices (?)

-- p