easy selectMap for simple features sf about mapedit

[timelyportfolio]

I first thought that the bug might be caused by wrapping a Shiny gadget in a function, but I cannot replicate the bug by simply wrapping in a function, so failure 1 on correcting.

library(mapview)
library(leaflet)
library(shiny)
library(mapedit)

sf_select <- function(xsf) {selectFeatures(xsf)}
#bsf <- sf_select(breweries91)

library(sf)
library(MODIS)
shp <- system.file("external", "modis_latlonWGS84_grid_world.shp", package = "MODIS")
grd <- st_read(shp)
gsf <- sf_select(grd)
plot(gsf)

but there might be a clue in this failure. If we uncomment the bsf <- ... line, the Shiny gadget only runs once when run inline.

[timelyportfolio]

@tim-salabim, should we use S3 and change selectFeatures() to selectMap.sf()? With resolution of r-spatial/mapview#85, will this become even easier?

[tim-salabim]

I am not sure. I feel that selectFeatures is more instructive as the intention is to select features not a map. Wouldn't it make sense to have selectMap for map objects (leaflet, mapview) and selectFeatures for spatial objects (sf, sp, ...)?

[timelyportfolio]

@fdetsch, as I look more deeply into the riddle of erratic return, I am even more confused since

x <- MODIS:::mapSelect()
str(x)

works without issue every time for me. I wonder if the error might be caused by some logic in getTile(), or if perhaps nesting the functions is the source of the trouble. I will keep researching.

[timelyportfolio]

I am starting to suspect dialogViewer as the potential source of our trouble. In my modules branch timelyportfolio@b6f0c17, I changed to use the default paneViewer instead of dialogViewer. After this change, I can no longer replicate the erratic getTile().

# devtools::install_github("timelyportfolio/mapedit@modules")

library(MODIS)
library(leaflet)
library(mapedit)

lapply(1:3, function(x) getTile())

@fdetsch, can you try the above on your end?

[timelyportfolio]

@tim-salabim, if this is the "solution", it will mean we will not be able to use the separate dialog popup. Instead, we will have to use the viewer pane or default browser. Is this acceptable? To see in action, run the following.

# devtools::install_github("timelyportfolio/mapedit@modules")
library(sf)
library(mapview)
library(mapedit)

lapply(1:3, function(x) selectFeatures(st_as_sf(franconia)))

I guess we could add a viewer argument that gets passed through to selectMap(). Once we decide how to handle this, we should probably make editMap consistent.

[tim-salabim]

@timelyportfolio this is acceptable. I agree that having a viewer argument would be good so we can control default (pane?) but user can choose to their liking.
I just had a play around and am very impressed! Even things like this work nicely:

leaflet() %>% 
  addTiles() %>% 
  addFeatures((sel <<- selectFeatures(franconia)), 
              popup = popupTable(sel)) %>% 
  addMouseCoordinates() %>%
  addLogo("https://avatars1.githubusercontent.com/u/837910?v=3&s=460")

I think this is nice as it enables people to flexibly set up their workflow.

[tim-salabim]

@timelyportfolio is it possible to make the map element in the viewer pane automatically resize to the available space?

[timelyportfolio]

Yes I think so but will verify this afternoon.

[timelyportfolio]

I thought I had figured this out at one point, but height="auto" and height="100%" both result in blank screens. Unfortunately, while width="100%" or width="auto" work as expected, we cannot do the same with height. I will try a different approach :(

[fdetsch]

Sorry for the late reply, guys. German public holidays and stuff...

@timelyportfolio, after adding some missing leaflet:: assignments to R/modules.R (see pull request #28), interactive tile selection through MODIS::getTile() seems to work permanently. Thanks for the great work!

[timelyportfolio]

I think we can close this. If anybody sees any issues, please reopen. Thanks all!

[tim-salabim]

@timelyportfolio any news on the issue of making the map element fit the available space in the paneViewer?

[timelyportfolio]

Will resume work on filling height and so that I don't forget again, file a new issue focused on this behavior.