basemap | map projections ( with coastlines and political boundaries | Data Visualization library

I have now got it working by using MarkerCluster.LayerSupport and by adding it to both var map and checking it in - maybe wrong plugin usage but likely my human error.

Working code below:

digging in the ipyleaflet code, it seems that the shadow is mandatory as it's only set in this css file. The different options are set in the js file meaning that shadow cannot be removed from python code.

As an ugly fix I forced some css directly on the top cell before import ipyleaflet:

The confusion came from a mis-understanding of how deck.gl's React component works & what those components really are (they're not react components).

Deck.gl's react component, DeckGL, intercepts all children and determines if they are in fact "layers masquerading as react elements" (see code). It then builds layers from each of those "elements" and passes them back to DeckGL's layers property.

They look like react components, but really aren't. They can't be rendered on their own in a React context. They can't be rendered outside of the DeckGL component at all, because they're still just plain deck.gl layers.

The solution here is to create a new map layer class just like you might in any other context (not a React component wrapping a layer). Docs for that are here.

You can achieve that by passing down a prop For instance

In this case you need to use an extent that includes all your geometries to initialize the view parameters, or after calculating zoom to that extent. For that you need to calculate the extent.

The particularity here is that your geometries are points, so you will not be able to use extent methods, because points have no extent.

But, not worries, to calculate the result extent (ie. the "full extent" of your geometries), is not difficult.

Here is a small function I put for you that can achieve that,

There are a couple of errors here.

1. apt cannot find libgeos

   E: Unable to locate package libgeos-3.5.0

   You may need to find another version of libgeos though I'm not sure what version. Perhaps this should match the installed dev version, 3.6.2?

2. pip cannot install using the archive that you are providing.

   ERROR: File "setup.py" not found for legacy project https://github.com/matplotlib/basemap/archive/master.zip.

   You can fix this by telling pip specifically what subdirectory it will find the package in:

Sorry for not directly responding to your described error, but I would not use esri-loader with newer versions of ArcGIS for JavaScript API. Why not npm as ES modules which do not require a separate script loader?

Build with ES Module

This way you can do simple imports like this:

import WebMap from "@arcgis/core/WebMap";

Here are the initial setup instructions:

Install instructions

Finally, here is a sample react app from Esri using exactly that:

Esri React App example

I don't know why it crashes for you, I didn't have this problem. Still, the main problem was that in app_server.R, you put map_id="basemap" when you call the module for itinerary.

In mod_basemap.R, the map is indeed called "basemap", but it is wrapped in ns(), which means that its actual id is "name you give when calling the module-" + "id you give to the input". Here, the actual id for the map is therefore "basemap_ui_1-basemap".

Now, it's not a good idea to specify this whole id (what if you replace "basemap_ui_1" by something else later?), so what you want is to return the map id when you call mod_basemap.R, so that you can use this id in other modules. So at the end of the server part in mod_basemap.R, you can add:

Rather looks like the Leaflet CSS is incorrectly loaded in your production bundle: tiles are scrambled up, no zoom and attribution controls.