@cldougl you might want to try http://nategood.com/sublime-text-strip-whitespace-save

to avoid those nasty diff lines.

will do

@alexcjohnson this has me thinking that perhaps flaglist attributes should be stored in fullData as arrays instead of strings, to make comparison more seamless. Moreover, maybe the all flag should be expected to e.g. x+y+text+name in the default step.

Alternatively, we could cook a nice general lib function isInFlagList that would take care of the logic on demand.

I like isInFlagList better (I like it a lot, in fact 👍 ) - fullData isn't supposed to change things from data unless they're invalid, it's just supposed to fill in the blanks.

Great. Let's keep that in mind when we get to #34. There are only a few flaglist attributes at the moment, but more are coming.

💥 nice <tspan>s

It looks like we could have 1 beforeEach and multiple its?
pseudocode

var mockData, mockLayout

describe('hover tests', function () {
   mockData = mockData()
   mockLayout = mockLayout()

   beforeEach( function (done) {
       Plotly.plot(gd, mockData, mockLayout).then(done)
   })

   it('responds to hover', function () { // test that uses mockData and mockLayout }
   it('resonds to hover x', function () { // test ... }
})

We can put the var declarations global scope and then assign fresh copies in the single BeforeEach. No need for multiple describes and beforeEach's. If that is dooable is should really clear up the test.

For comparison check out the many Jest (Jasmine fork) tests in Filewell. We have a set of patterns that might be useful. https://github.com/plotly/streambed/blob/master/shelly/filewell/static/filewell/src/stores/__tests__/DirectoryStore-test.js

Here there is one BeforeEach to refresh state before each test and a few describe blocks that break the test into logical chunks. If we are just testing hover here one describe is probably enough

Check around the filewell src tree, every folder has a __test__ folder with Jest tests.

Nop. That won't because we call Plotly.plot with different data each time.

We could only force update hovermode in gd directly without a plot call and test the labels then. But that feels kind of dirty.

ah aight - the only other suggestion I can think of is to remove the multiple BeforeEach and Describes and do something like

it('does hover things', function(done) {
    Plotly.plot(newGD(), newData(), newLayout()).then( function () {
       expect(thingsToHappen()).toBe(true)
       done()
    })
})

where you get shorter code but have 1 level of nesting in the tests.