Snapshot not recognized for sqn_order fold?

[ThomasArts]

The following test case is suspicious, because even though we fold with snapshot true we don't seem to get the keys from only the snapshot.

leveledjc_eqc:init_backend(o_rkv,
    [{root_path, Dir}, {log_level, error}, {max_sstslots, 2},
     {cache_size, 10}, {max_pencillercachesize, 40},
     {max_journalsize, 20000}],
    sut) ->
  <0.2667.0>
leveledjc_eqc:put(<0.2667.0>, <<98, 117, 99, 107, 101, 116, 51>>,
    <<227, 142, 0, 237, 130, 199, 37, 238, 62, 225, 139, 160, 158,
      68, 105, 142>>,
    <<53, 1, 0, 0, 0, 43, 131,....>>,
    [], o_rkv) ->
  ok

Normal start, putting 1 object. Note that object has 43 and 131 in it to distinguish it later on.
Then:

leveledjc_eqc:objectfold(<0.2667.0>, o_rkv, {#Fun<leveledjc_eqc.30.5533705>, []},
    true, sqn_order, 4) ->
  #Fun<leveled_runner.10.98249213>
leveledjc_eqc:put(<0.2667.0>, <<98, 117, 99, 107, 101, 116, 51>>,
    <<227, 142, 0, 237, 130, 199, 37, 238, 62, 225, 139, 160, 158,
      68, 105, 142>>,
    <<53, 1, 0, 0, 0, 78, 131,...>>, [], o_rkv) -> ok

Start an object fold in sequence order with SNAPSHOT true! After that, with same bucket and key add a new object. Note the 78, 131 there.
If we now run the fold, the implementation returns:

leveledjc_eqc:fold_run(4, #Fun<leveled_runner.10.98249213>) ->
  [{<<98, 117, 99, 107, 101, 116, 51>>,
    <<227, 142, 0, 237, 130, 199, 37, 238, 62, 225, 139, 160, 158,
      68, 105, 142>>,
    <<53, 1, 0, 0, 0, 78, 131,...>>},
    {<<98, 117, 99, 107, 101, 116, 51>>,
    <<227, 142, 0, 237, 130, 199, 37, 238, 62, 225, 139, 160, 158,
      68, 105, 142>>,
    <<53, 1, 0, 0, 0, 43, 131, ...>>}]

Thus both objects. This surprises me a bit, I had expected to only get the object back from the snapshot moment.

[martinsumner]

martinsumner#261

There is an eunit test leveled_bookie:sqnorder_mutatefold_test/0 that recreates the problem, and now passes after the change.