Internal data by axis instead of scale id

[kurkle]

Relevant part of parseObjectData after babel:
pr

    for (i = 0, ilen = count; i < ilen; ++i) {
      index = i + start;
      item = data[index];
      parsed[i] = {
        x: xScale._parseObject(item, 'x', index),
        y: yScale._parseObject(item, 'y', index)
      };
    }

Locally run tests:

100 runs done in 15430ms. Average: 154ms, min: 92ms, max: 516ms, variation: 424ms
100 runs done in 16681ms. Average: 166ms, min: 87ms, max: 867ms, variation: 780ms

In plunker:

50 runs done in 6431ms. Average: 128ms, min: 87ms, max: 441ms, variation: 354ms

master

    for (i = 0, ilen = count; i < ilen; ++i) {
      var _parsed$i3;

      index = i + start;
      item = data[index];
      parsed[i] = (_parsed$i3 = {}, _defineProperty(_parsed$i3, xId, xScale._parseObject(item, 'x', index)), _defineProperty(_parsed$i3, yId, yScale._parseObject(item, 'y', index)), _parsed$i3);
    }

Locally run tests:

100 runs done in 19970ms. Average: 199ms, min: 118ms, max: 539ms, variation: 421ms
100 runs done in 22556ms. Average: 225ms, min: 109ms, max: 863ms, variation: 754ms

In plunker:

50 runs done in 8775ms. Average: 175ms, min: 118ms, max: 461ms, variation: 343ms

plunker

(all tests run on chrome 79)

[etimberg]

I didn’t see anything wrong. Just to be safe though, can we test with a chart that has extra x and y axes? Since the default axis IDs are now x and y I want to be sure we got all the spots

[benmccann]

If you have two y scales, y1 and y2, won't they both get parsed to y and the second one will override the data from the first?

[kurkle]

The parsed is per dataset (unless parsing is disabled and supplying same data to multiple datasets)
(and a dataset is obviously bound to single y-scale)
The parsing: false case suffers from this, you can't have data like {time: 123414, line: 10, bar: 15} and just give the scales id:s 'time', 'line' and 'bar'

[kurkle]

I'll check re-check docs etc on this, lets not merge yet.

[kurkle]

Ok, ready for another review