geoscripting-wur / scripting4geo Goto Github PK

I'm not sure whether it's intentional that the R code for the second interactive map demo (https://github.com/GeoScripting-WUR/Scripting4Geo/blob/gh-pages/index.Rmd#demo-2-interactive-maps) is not printed (echo=FALSE). It seems a bit odd to have the code working but not show it?

I will work on that today.
Any other ideas - on recent nice R geo-spatial packages?

In https://github.com/GeoScripting-WUR/Scripting4Geo/blob/gh-pages/index.Rmd#demo-integrated-analysis-chain-from-download-to-analysis this line:

out <- extract(bel, Sint, df = TRUE, sp = TRUE)

fails with this error:

Error in SpatialPointsDataFrame(y, v[, -1, drop = FALSE]) : 
  row.names of data and dimnames of coords do not match
In addition: Warning message:
In SpatialPointsDataFrame(y, v[, -1, drop = FALSE]) :
  forming a SpatialPointsDataFrame based on maching IDs, not on record order. Use match.ID = FALSE to match on record order
Error in SpatialPointsDataFrame(y, v[, -1, drop = FALSE]) : 
  row.names of data and dimnames of coords do not match

I tested it on the VM (R 3.2.3) as well as my own machine (R 3.3.0) and the error is the same.

This is a rather strange issue, considering that sp=FALSE works.

are there github based options for a q/a or discussion platform for when the students are stuck during the tutorials during the GeoScripting-WUR course?

Right now the vector handling in this lesson is still done with sp. It should be updated with sf instead.

On the VM, download.file() doesn't run properly without method="curl" or method="wget" (the help file suggests the former). Of course, it's also platform-dependent; I'm not sure if curl is available on Windows. Maybe checking for the platform each time would be a solution... Or perhaps making use of another package for downloading files.

the style is nice so can be applied to other repositories.

Some students were confounded when seeing the code in this lesson, as it's not explained and quite complicated. It should be clearly stated that it involves raster and vector manipulation that is explained in upcoming lessons, and that they are not yet supposed to know exactly how and why this code works.

Andrei said that the problem why download.file() would always fail on Windows is because it assumes text and not binary. So the solution is to always explicitly set mode to be wb. We should do that for all downloads, then.

Looks like regular rmarkdown, when generating the HTML file, includes the entirety of htmlwidgets and Leaflet JavaScript libraries, and the output file is about 1 MiB in size. knitrBootstrap on the other hand does not, the output is ~25 KiB. The framework for the map is generated, but there are no libraries to actually act upon the HTML objects placed at the spot we want to have the map.

Just including Leaflet is also not enough, and htmlwidgets don't seem to have a hosted version, so just editing the output file doesn't seem to be an option.

The shiny example is included using an iframe, perhaps as a workaround the leaflet map can also be included as an iframe?