Uh oh!
There was an error while loading. Please reload this page.
Multiple scattergeo improvement - #1004
Conversation
- which is consistent with mapbox subplot, and comptabitible with Fx.hover
- to be reuse in 'scattergeo'
- which skips and cast lon/lat values - fixes#963 - note that 'locations[i]' -> feature must still be done at the plot (after the topojson is loaded)
- traces with no data points are given a place-holder calc trace to retatin fullData.length === gd.calcdata.length, and make some component (e.g. legend) logic easier - tag these place-holder traces so that the plot modules can easily skip over them
- so plotting that data-pt less traces do result in mapbox errors.
- add support for 'connectgaps' - add support for 'fill: 'toself' - remove all traces of hover / click handlers
- make Fx.hover pass subplot info to hoverPoints module - add scattergeo hoverPoints and eventData modules - mock xaxis and yaxis in geo instances - factor out is-over-edge logic to use it in hoverPoints - use 'mousemove' instead of 'mouseover' in test to trigger hover
- until we make choropleth use fullLayout._hoverlayer for its hover labels.
- the first legend item is now correct!
bpostlethwaite
left a comment
There was a problem hiding this comment.
Looks ok to me but I didn't get right into the details for all these changes. If I didn't understand what the code was doing fairly quickly I made a comment about improving the naming or comments.
| ya = pointData.ya, | ||
| geo = pointData.subplot; | ||
| if(cd[0].placeholder) return; |
There was a problem hiding this comment.
this is a little mysterious, at least as someone who doesn't look at Plotly.js code daily. Maybe a comment here?
There was a problem hiding this comment.
Here's the comment you're looking for: https://github.com/plotly/plotly.js/pull/1004/files#diff-ad4f76ccd6044ed16514297078e13b84R1672
Putting comment about something where it is set is probably more robust than putting it where it is used.
| var lonlat = d.lonlat; | ||
| // this handles the not-found location feature case | ||
| if(lonlat[0] === null || lonlat[1] === null) return Infinity; |
There was a problem hiding this comment.
why return Infinity and not say, null
There was a problem hiding this comment.
because this function returns a distance.
There was a problem hiding this comment.
I would have expected NaN I think.
| var di = cd[pointData.index], | ||
| lonlat = di.lonlat, | ||
| pos = c2p(lonlat), | ||
| rad = di.mrc || 1; |
There was a problem hiding this comment.
is di.mrc undefined or 0? Or could be both and we want it to be 1 regardless. Also why 1? Maybe some comments here
There was a problem hiding this comment.
di.mrc is set only when there mode includes 'markers', so we need a fallback for mode 'line' and 'text'
There was a problem hiding this comment.
I wish I could gather that from that line of code ;)
| var dx = Math.abs(xPx - pos[0]), | ||
| dy = Math.abs(yPx - pos[1]), | ||
| rad = Math.max(3, d.mrc || 0); |
There was a problem hiding this comment.
what are the constraints on radius here for?
There was a problem hiding this comment.
marker pt radii are taken into consideration in the picking routine.
There was a problem hiding this comment.
Do you think it will be obvious to someone who is maintaining this code what the 3 and 0 are doing? I didn't get it right away which is usually a sign that a comment may be helpful
| if(!lonlat) continue; // filter the blank points here | ||
| // skip over placeholder traces | ||
| if(calcTrace[0].placeholder) s.remove(); |
| xa = pointData.xa, | ||
| ya = pointData.ya; | ||
| if(cd[0].placeholder) return; |
There was a problem hiding this comment.
is there another name we can use beside placeholder, a placeholder for _____? How about emptytrace?.
There was a problem hiding this comment.
we could.
Not sure emptytrace is better though. traceWithNoVisibleDataPoint would be the most verbose. But I think placeholder does the trick as discussed in https://github.com/plotly/plotly.js/pull/1004/files#diff-ad4f76ccd6044ed16514297078e13b84R1672
There was a problem hiding this comment.
yep placeholder is ok. Perhaps we can expand a little in that original placeholder comment. I still didn't quite understand when they would be necessary
bpostlethwaite
commented
Oct 5, 2016
Aighty all the comments are nonblocking! There are nice tests and this PR looks like a huge win Nice work! |
bpostlethwaite
commented
Oct 5, 2016
💃 |
fixes#963
and supersedes #964
In brief, this PR brings several ideas from
scattermapboxover toscattergeo- while reusing some parts of logic (namely regarding GeoJSON) keeping things 🌴In details, this PR:
scattergeouse a proper calc stepscattergeouseFx.hoverto pick (using calcdata) and draw its hover labelsmode: 'lines'now (I mean finally) has a hover handlers!connecgapsto scattergeofill: 'toself'to scattergeo (similar to howscattermapboximplementfill)🎉 Yay geo line hover text: