Skip over placeholder calc traces in bar set positions

[etpinard]

fixes#1310

In bar calc, bars with non-numeric positions are not included in the resulting gd.calcdata (ref). Later in Plots.doCalcdata, those skipped bars are given a placeholder calcdata item (ref) to ensure that all calc traces have as many items their corresponding fullData traces (see dd05c6e for more on placeholder calcdata items).

These placeholder calc items contain x: false, y: false, which throws off the bar min-difference here; hence the 🐛 . In details, false is coerced to 0 in Lib.distinctVals so essentially

data=[{x: [1,2,3],y: [2,1,2]},{x: [null],y: [/* ... */]}]

was treated essentially the same way as

data=[{x: [1,2,3],y: [2,1,2]},{x: [0],y: [/* ... */]}]

which leads to large errors especially when data[0] has date positions (which are stored in datetimes at the set position step) as reported in #1310.

In brief, the fix is easy, simply skip over placeholder calc traces during bar set positions.

[alexcjohnson]

💃

[etpinard]

fixes#1310

In bar calc, bars with non-numeric positions are not included in the resulting gd.calcdata (ref). Later in Plots.doCalcdata, those skipped bars are given a placeholder calcdata item (ref) to ensure that all calc traces have as many items their corresponding fullData traces (see dd05c6e for more on placeholder calcdata items).

These placeholder calc items contain x: false, y: false, which throws off the bar min-difference here; hence the 🐛 . In details, false is coerced to 0 in Lib.distinctVals so essentially

data=[{x: [1,2,3],y: [2,1,2]},{x: [null],y: [/* ... */]}]

was treated essentially the same way as

data=[{x: [1,2,3],y: [2,1,2]},{x: [0],y: [/* ... */]}]

which leads to large errors especially when data[0] has date positions (which are stored in datetimes at the set position step) as reported in #1310.

In brief, the fix is easy, simply skip over placeholder calc traces during bar set positions.

[alexcjohnson]

💃

[etpinard]

fixes#1310

In bar calc, bars with non-numeric positions are not included in the resulting gd.calcdata (ref). Later in Plots.doCalcdata, those skipped bars are given a placeholder calcdata item (ref) to ensure that all calc traces have as many items their corresponding fullData traces (see dd05c6e for more on placeholder calcdata items).

These placeholder calc items contain x: false, y: false, which throws off the bar min-difference here; hence the 🐛 . In details, false is coerced to 0 in Lib.distinctVals so essentially

data=[{x: [1,2,3],y: [2,1,2]},{x: [null],y: [/* ... */]}]

was treated essentially the same way as

data=[{x: [1,2,3],y: [2,1,2]},{x: [0],y: [/* ... */]}]

which leads to large errors especially when data[0] has date positions (which are stored in datetimes at the set position step) as reported in #1310.

In brief, the fix is easy, simply skip over placeholder calc traces during bar set positions.

[alexcjohnson]

💃

[etpinard]

fixes#1310

In bar calc, bars with non-numeric positions are not included in the resulting gd.calcdata (ref). Later in Plots.doCalcdata, those skipped bars are given a placeholder calcdata item (ref) to ensure that all calc traces have as many items their corresponding fullData traces (see dd05c6e for more on placeholder calcdata items).

These placeholder calc items contain x: false, y: false, which throws off the bar min-difference here; hence the 🐛 . In details, false is coerced to 0 in Lib.distinctVals so essentially

data=[{x: [1,2,3],y: [2,1,2]},{x: [null],y: [/* ... */]}]

was treated essentially the same way as

data=[{x: [1,2,3],y: [2,1,2]},{x: [0],y: [/* ... */]}]

which leads to large errors especially when data[0] has date positions (which are stored in datetimes at the set position step) as reported in #1310.

In brief, the fix is easy, simply skip over placeholder calc traces during bar set positions.

[alexcjohnson]

💃

[etpinard]

fixes#1310

In bar calc, bars with non-numeric positions are not included in the resulting gd.calcdata (ref). Later in Plots.doCalcdata, those skipped bars are given a placeholder calcdata item (ref) to ensure that all calc traces have as many items their corresponding fullData traces (see dd05c6e for more on placeholder calcdata items).

These placeholder calc items contain x: false, y: false, which throws off the bar min-difference here; hence the 🐛 . In details, false is coerced to 0 in Lib.distinctVals so essentially

data=[{x: [1,2,3],y: [2,1,2]},{x: [null],y: [/* ... */]}]

was treated essentially the same way as

data=[{x: [1,2,3],y: [2,1,2]},{x: [0],y: [/* ... */]}]

which leads to large errors especially when data[0] has date positions (which are stored in datetimes at the set position step) as reported in #1310.

In brief, the fix is easy, simply skip over placeholder calc traces during bar set positions.

[alexcjohnson]

💃

[etpinard]

fixes#1310

In bar calc, bars with non-numeric positions are not included in the resulting gd.calcdata (ref). Later in Plots.doCalcdata, those skipped bars are given a placeholder calcdata item (ref) to ensure that all calc traces have as many items their corresponding fullData traces (see dd05c6e for more on placeholder calcdata items).

These placeholder calc items contain x: false, y: false, which throws off the bar min-difference here; hence the 🐛 . In details, false is coerced to 0 in Lib.distinctVals so essentially

data=[{x: [1,2,3],y: [2,1,2]},{x: [null],y: [/* ... */]}]

was treated essentially the same way as

data=[{x: [1,2,3],y: [2,1,2]},{x: [0],y: [/* ... */]}]

which leads to large errors especially when data[0] has date positions (which are stored in datetimes at the set position step) as reported in #1310.

In brief, the fix is easy, simply skip over placeholder calc traces during bar set positions.

[alexcjohnson]

💃

[etpinard]

fixes#1310

In bar calc, bars with non-numeric positions are not included in the resulting gd.calcdata (ref). Later in Plots.doCalcdata, those skipped bars are given a placeholder calcdata item (ref) to ensure that all calc traces have as many items their corresponding fullData traces (see dd05c6e for more on placeholder calcdata items).

These placeholder calc items contain x: false, y: false, which throws off the bar min-difference here; hence the 🐛 . In details, false is coerced to 0 in Lib.distinctVals so essentially

data=[{x: [1,2,3],y: [2,1,2]},{x: [null],y: [/* ... */]}]

was treated essentially the same way as

data=[{x: [1,2,3],y: [2,1,2]},{x: [0],y: [/* ... */]}]

which leads to large errors especially when data[0] has date positions (which are stored in datetimes at the set position step) as reported in #1310.

In brief, the fix is easy, simply skip over placeholder calc traces during bar set positions.

[alexcjohnson]

💃

[etpinard]

fixes#1310

In bar calc, bars with non-numeric positions are not included in the resulting gd.calcdata (ref). Later in Plots.doCalcdata, those skipped bars are given a placeholder calcdata item (ref) to ensure that all calc traces have as many items their corresponding fullData traces (see dd05c6e for more on placeholder calcdata items).

These placeholder calc items contain x: false, y: false, which throws off the bar min-difference here; hence the 🐛 . In details, false is coerced to 0 in Lib.distinctVals so essentially

data=[{x: [1,2,3],y: [2,1,2]},{x: [null],y: [/* ... */]}]

was treated essentially the same way as

data=[{x: [1,2,3],y: [2,1,2]},{x: [0],y: [/* ... */]}]

which leads to large errors especially when data[0] has date positions (which are stored in datetimes at the set position step) as reported in #1310.

In brief, the fix is easy, simply skip over placeholder calc traces during bar set positions.

[alexcjohnson]

💃