Key tree property tests

[nickva]

Key tree module is a candidate to use property tests on as it mostly deals with
manipulating a single data structure and functions are referentially
transparent, that is, they aren't many side-effects like IO for example.

The test consists of two main parts - generators and properties.

Generators generate random input, for example revision trees, and properties
check that certain properties hold, for example that after stemming all the
leaves are still present in the revtree.

To run the test:

make eunit apps=couch suites=couch_key_tree_prop_tests

• Code is written and works correctly;
• Changes are covered by tests;
• Documentation reflects the changes;

[wohali]

Need a review from someone more familiar with this code than I, I'm not willing to +1 this on my own.

Also needs the conflict resolved in rebar.config.script.

[eiri]

I am willing to +1 it, the tests themselves look good, but they, understandably, take a long time to run. I just re-run just that one suite on my not too weak MBP and it took [done in 287.081 s], so I'm afraid it'll kick Travis and Jenkins in the guts if it'd be part of standard eunit run.

@nickva do you think we can move it to a separate make target somehow? It's nice to have those, don't get me wrong, but they are quite specific bit, so we probably don't need to run them for every change?

[nickva]

@eiri @wohali Thanks for taking a look

I rebased it. I'll see maybe I'll try to reduce the branchiness and the generated sizes a bit to speed things up. So maybe it adds another 2-3 minutes but not 5

How are these times looking?

module 'couch_key_tree_prop_tests'
  couch_key_tree_prop_tests:49: revtree_merge_random_nodes_test_...[0.493 s] ok
  couch_key_tree_prop_tests:154: after_stemming_paths_are_shorter_test_...[0.269 s] ok
  couch_key_tree_prop_tests:175: leaf_count_test_...[0.255 s] ok
  couch_key_tree_prop_tests:136: after_stemming_all_leaves_are_present_test_...[0.262 s] ok
  couch_key_tree_prop_tests:99: stemming_results_in_same_or_less_total_revs_test_...[0.260 s] ok
  couch_key_tree_prop_tests:63: revtree_merge_some_existing_some_new_test_...[23.715 s] ok
  couch_key_tree_prop_tests:117: stem_path_expect_size_to_get_smaller_test_...[5.431 s] ok
  couch_key_tree_prop_tests:182: get_leafs_test_...[0.224 s] ok
  couch_key_tree_prop_tests:32: revtree_merge_with_subset_of_own_nodes_test_...[35.110 s] ok
  couch_key_tree_prop_tests:84: no_change_stemming_deeper_than_current_depth_test_...[0.253 s] ok
  [done in 66.302 s]

[eiri]

Ok, [done in 67.107 s] locally and ~30mins on Travis, I think we can live with that. Merge it and we'll see what Jenkins have to say.

[wohali]

For posterity, we need to mirror new dependencies into Apache infrastructure when adding them - not just check that they are compatibly licensed. This should have been done at the same time as this merge landing.