some datapoint values in the response are wrong

[buchslava]

dataset is SG
query is

{
        select: {
          key: ['geo', 'time'],
          value: [
            'life_expectancy_years', 'population_total'
          ]
        },
        from: 'datapoints',
        where: {
          $and: [
            {geo: '$geo'},
            {time: '$time'},
            {
              $or: [
                {population_total: {$gt: 10000}},
                {life_expectancy_years: {$gt: 30, $lt: 70}}
              ]
            }
          ]
        },
        join: {
          $geo: {
            key: 'geo',
            where: {
              $and: [
                {'is--country': true},
                {latitude: {$lte: 0}}
              ]
            }
          },
          $time: {
            key: 'time',
            where: {$and: [{time: {$gt: '1990', $lte: '2015'}}]}
          }
        },
        order_by: ['time', 'geo']
      }

or

http://waffle-server-dev.gapminderdev.org/api/ddf/ql?_select_key@=geo&=time;&value@=life/_expectancy/_years&=population/_total;;&from=datapoints&where_$and@_geo=$geo;&_time=$time;&_$or@_population/_total_$gt:10000;;&_life/_expectancy/_years_$gt:30&$lt:70;;;;;;&join_$geo_key=geo&where_$and@_is--country:true;&_latitude_$lte:0;;;;;&$time_key=time&where_$and@_time_$gt=1990&$lte=2015;;;;;;&order/_by@=time&=geo;&dataset=buchslava%252Freaders-test-ds-systema-globalis

result is (see ws.json)

life_expectancy_years values should not be null

[buchslava]

The reason for the issue is due to WS functionality: WS does not support measure filter applying for the row. It supports filter applying for column (separate measure) only. Let me explain...

Imagine, we have next filter and next result for a dataset (datapoints selection):
Filter is empty {}
Result is

geo | time  | life_expectancy | population
arg  | 1991 | 72                      | 900000

An another case for same dataset:
Filter is { $or: [ {population_total: {$gt: 10000}}, {life_expectancy_years: {$gt: 30, $lt: 70}} ] }

Result for DDFCSV reader is

geo | time | life_expectancy | population
arg | 1991 | 72 | 900000

because 72>70 is false OR 90000>10000 is true -> result is true

Result for WS reader is

geo | time | life_expectancy | population
arg | 1991 | null                     | 900000

because WS applies the filter for each measure separately and merges result after that

I guess it's incorrect in accordance with https://docs.google.com/document/d/1olFm-XXjWxQ4LrTCfM42an6LbjbIgnt__V1DZxSmnuQ/edit# (Filter which affects which rows of the selection will be included in the response. This filter is a subset of MongoDB query operators)