When we calculate the amount of missing data, in particular with the synonym support I'm about to pull-request, we should also make some tests for that.
I notice it because I'm currently looking at two versions of counting missing data, and I'm somewhat sure that the second one is the one I want, but a few explicit tests would give me rest in this issue.
Tests to write could have 3 languages with 2 features each, with various combinations:
[ ] each of the six slots has exactly one value
[ ] 5 normal values, ? in 6th
[ ] 5 normal values, - in 6th
[ ] 5 normal values, `` in 6th
[ ] 5 normal values, 6th missing, not listed
[ ] 5 normal values, 6th containing two different values
[ ] 5 normal values, 6th contains both a value and a gap character
[ ] L1 has a value for three features each, L2 a value for two features, L3 a value for one feature.
When we calculate the amount of missing data, in particular with the synonym support I'm about to pull-request, we should also make some tests for that.
I notice it because I'm currently looking at two versions of counting missing data, and I'm somewhat sure that the second one is the one I want, but a few explicit tests would give me rest in this issue.
Tests to write could have 3 languages with 2 features each, with various combinations:
?in 6th-in 6thAnything else?