Transformed array on zoom fix #1717
Conversation
- flag fullData traces that passed through calc transforms with `_hasCalcTransform: true` - update calcdata[i][0].trace ref BUT map back transformed arrays so that they match the calcdata[i][j] items on non-recalc updates (e.g. zoom events)
| @@ -505,12 +506,38 @@ plots.supplyDefaults = function(gd) { | |||
| // update object references in calcdata | |||
| if((gd.calcdata || []).length === newFullData.length) { | |||
There was a problem hiding this comment.
This block updates the gd.calcdata[i][0].trace reference so that the plot step can use the correct fullData attribute values even in cases where Plots.doCalcdata isn't called (e.g. on zoom).
src/plots/plots.js
Outdated
|
|
||
| for(i = 0; i < arrayAttrs.length; i++) { | ||
| ast = arrayAttrs[i]; | ||
| Lib.nestedProperty(cd0.trace, ast).set(transformedArrayHash[ast]); |
There was a problem hiding this comment.
Here we make sure that the array attributes in gd.calcdata[i][0].trace match the transformed array post calc transform. Other non-data-array attributes are left untouched.
src/plots/plots.js
Outdated
| var oldTrace = cd0.trace; | ||
| var arrayAttrs = PlotSchema.findArrayAttributes(oldTrace); | ||
| var transformedArrayHash = {}; | ||
| var i, ast; |
There was a problem hiding this comment.
🐄 just so I don't get confused, ast makes me think "abstract syntax tree", elsewhere we call this astr
There was a problem hiding this comment.
Oh right, astr is what I had in mind. Something wrong happened between brain and keyboard.
|
Nice use of Yeah, annoying that we have to do this (for the record from a separate chat: the underlying issue is that a 💃 after #1717 (comment) |
|
Is it clear what needs to happen with this? I'm still having just a bit of trouble understanding exactly why this is needed and then the downstream problem with animation. |
First look at the added assertions. If that's not clear, slack me. |
fixes #1713
As promised, it's not pretty. cc @n-riesco @alexcjohnson