mapbox / guidance-replay Goto Github PK

Currently, the getSteps function for v4 responses matches the coordinates (1) for each maneuver to (2) the coordinates the comprise the LineString, and use those coordinates to create the step feature.

I'm running into the issue where some maneuver coordinates have a different number of decimal places than the coordinates in the route geometry (example where maneuver latitudes have 6 decimal places and route geometry latitudes have 5).

@yhahn I'll poke at adding some catches here with some additional tests and see if that helps fix some of the problems with using v4 routes in the simulator. Sound bueno?

Not sure where it'll fit into our suite of tools long run but parking here is fine for now.

Cutting a ticket so we don't forget that there may be a better home for this class elsewhere 👌

As of now, I'm using time to keep track of what step we're on in the route. Current step is a required parameter for integration with navigation.js.

This works great when "spacing": "constant", but the instructions start to behave unexpectedly around maneuvers. I think this is because we formulaically ease from the speed of the current segment into the speed of the next segment, which offsets our current time & coordinate combination from the expectation. At maneuvers, we are farther along the route than is expected by the function.

I'm trying to think of a solution that better integrates with "spacing": "acceldecel" mode. I could reattempt distance-based step identification, but had ultimately decided on time-based identification after struggling with highway cloverleaf interchanges. I wonder if there's a way to pull in speedChange into the Locator to better prepare for maneuvers.

cc @yhahn @emilymcafee