project-gen3sis/r-package

Repository of the R-packageGen3sis

r-package's Issues

starting richness plots and range plots(pdfs) are not the same as when called for the first timestep in the call_observer function.

South America distances missing

Impossible to follow the tutorial, since distances matrices are currently not provided with the package. Either provide a line in the vignette to compile these distances or use another landscape as example at the vignette

multiple richenss plot with tag -1 when simulation starts with multiple species

Reproduce issue:
datapath <- system.file(file.path("extdata", "SouthAmerica"), package="gen3sis")
config_object_old <- create_input_config(file.path(datapath, "config/config_southamerica.R"))
sim_old <- run_simulation(config = config_object_old,
landscape = file.path(datapath, "landscape"),
verbose=0, # no progress printed
output_directory=tempdir())

Example files

Example files are running too long for CRAN - shorten or wrap in \donttest{}, see https://cran.r-project.org/web/packages/roxygen2/vignettes/rd.html

Examples with CPU or elapsed time > 5s
                 user system elapsed
run_simulation 32.379   0.58  33.844

precompute partial distance matrices

Currently the distance matrices can only be left out or computed fully during the input creation.

We may want to offer calculating partial/full distance matrices from existing inputs.

add to imports

• checking R code for possible problems ... NOTE
  conditional_plot : : no visible global function definition
  for 'is'
  conditional_plot: no visible global function definition for 'png'
  Undefined global functions or variables:
  is png
  Consider adding
  importFrom("grDevices", "png")
  importFrom("methods", "is")
  to your NAMESPACE file (and ensure that your DESCRIPTION Imports field
  contains 'methods').

CRAN Fixes gen3sis 1.4

##1
If there are references describing the methods in your package, please add these in the description field of your DESCRIPTION file in the form authors (year) doi:... authors (year) arXiv:... authors (year, ISBN:...) or if those are not available: https:... with no space after 'doi:', 'arXiv:', 'https:' and angle brackets for auto-linking.
(If you want to add a title as well please put it in quotes: "Title")

##2
Please add \value to .Rd files regarding exported methods and explain the functions results in the documentation. Please write about the structure of the output (class) and also what the output means. (If a function does not return a value, please document that too, e.g.
\value{No return value, called for side effects} or similar) Missing Rd-tags in up to 18 .Rd files, e.g.:
evolution_mode_none.Rd: \value
plot_landscape_overview.Rd: \value
plot_landscape.Rd: \value
plot_ranges.Rd: \value
plot_raster_multiple.Rd: \value
plot_raster_single.Rd: \value
...

##3
Please always make sure to reset to user's options(), working directory or par() after you changed it in examples and vignettes and demos.
e.g.: inst/doc/introduction.R
oldpar <- par(mfrow = c(1,2))
...
par(oldpar)

Please do not modify the .GlobalEnv. This is not allowed by the CRAN policies. e.g.: .Random.seed

##4
Additionally:
Have the issues why your package was archived been fixed?
Please explain this in the submission comments.

plot_summary compatible with plot_richness

inhabited sites should appear in a different color scale in cases where 0 %in% richness

landscapes argument of the create_input()

So, I tried to use for the function's argument the list of data frames that I used to produce the rds file before. Then I get this error "unkown landscapes; it has to be a named list of list of either rasters or raster files..."
This means I need to transform my data frames into rasters instead of directly providing a list of data frames that can be directly saved into rds. That seems a bit counterproductive. It would be nice if the function would recognize that the list of landscapes is already in the format necessary to write a .rds.
Meanwhile I have to check how to transform my arrays into raster stacks then so the function can use them.

use only one config parameter not config file and config object

Setting to save plots to file?

either by config flag or switch on interactive()?

plot_richness on vignet not workin after save functionality added

the plot_richness in the introduction vignette are now showing up anymore
(see lines 233-241 on introduction.Rmd)
I think it has to do with you changes

specificy alternatives on plot_summary documentaiton regarding the sgen3sis.rds

Why so many vignettes?

@ohagen - is there are special reason why we wouldn't put everything in the same vignette?

calling of plots from observer not working

If you add the following lines to the observer function:

plot_richness(species, landscape)
plot_landscape(landscape)
plot_species_presence(species[[1]], landscape)"

run

run_simulation(config = "./inst/extdata/WorldCenter/config/config_worldcenter.R", 
+                     landscape = "./inst/extdata/WorldCenter/landscape",
+                     call_observer = "all", output_directory = "YOURTEMPDIRHERE")

breaks with error:

[1] "config found: ./inst/extdata/WorldCenter/config/config_worldcenter.R"
[1] "landscape found: ./inst/extdata/WorldCenter/landscape"
[1] "output directory is: C:/temp/config_worldcenter"
Error in rownames(landscape[["coordinates"]]) : 
  object 'landscape' not found

implement CI

previously relying on Travis, now we need another solution. Specially if we want to improve collaboration

Modify write_phylogeny so that nex object is correct toughout the simulation run

Vignettes cleanup

Hi Oskar,

I just looked at the vignettes because of the CRAN check problems - if you want to push this on CRAN soon, I would clean up the vignettes before. If you have the money, this could also be a work that a student helper with good english could do.

Build error due to create_input.Rmd

Hi Juliano, this file currently doesn't run, probably due to #24

No problem, but now that we have the CI running, to be able to check the model, it would be better move this file into a new branch, so that we have the master branch clean / can check if everything works fine.

I can do it for you, I just didn't want to do it in case you are currently working on it.

axis label for time series in summary plot is wrong

the indexing for the timestep labels on the x-axis is wrong.

line 132-134

  axis_lab <- seq(from=1, to=nrow(d), length.out = max((nrow(d)/20),2))
  axis(1, at=rev(as.numeric(rownames(d)))[axis_lab], labels = rownames(d)[axis_lab])
  mtext(side=1, text="Time steps", line=2.5, cex=1.1)

the "at" uses the row number 0 as starting index. indexing is positional and starts at 1. hence all axis labels are shifted by one to the left

compile input landscape from dataframs or rds as well, not only from rasters

should be present on v1.1

distance matrix test fails now

I suspect that this could have something with the new change of the distances in Km to m (1/1000)
When checking -as --cran on R.4.0:

*** checking for unstated dependencies in 'tests' ... OK
* checking tests ...
** running tests for arch 'i386' ...
  Running 'testthat.R'
 ERROR
Running the tests in 'tests/testthat.R' failed.
Last 13 lines of output:
  > 
  > test_check("gen3sis")
  -- 1. Failure: distance matrix creation works (@test-distances.R#26)  ---------------------------------------------------------------------------------------
  isTRUE(all.equal(unname(dist_m), gdist_m)) isn't true.
  
  -- 2. Failure: get_local_distances works (@test-input_creation.R#134)  --------------------------------------------------------------------------------------
  isTRUE(all.equal(unname(distance_local), local_tr)) isn't true.
  
  == testthat results  ========================================================================================================================================
  [ OK: 26 | SKIPPED: 10 | WARNINGS: 0 | FAILED: 2 ]
  1. Failure: distance matrix creation works (@test-distances.R#26) 
  2. Failure: get_local_distances works (@test-input_creation.R#134) 
  
  Error: testthat unit tests failed
  Execution halted
** running tests for arch 'x64' ...
  Running 'testthat.R'
 ERROR
Running the tests in 'tests/testthat.R' failed.
Last 13 lines of output:
  > 
  > test_check("gen3sis")
  -- 1. Failure: distance matrix creation works (@test-distances.R#26)  ---------------------------------------------------------------------------------------
  isTRUE(all.equal(unname(dist_m), gdist_m)) isn't true.
  
  -- 2. Failure: get_local_distances works (@test-input_creation.R#134)  --------------------------------------------------------------------------------------
  isTRUE(all.equal(unname(distance_local), local_tr)) isn't true.
  
  == testthat results  ========================================================================================================================================
  [ OK: 26 | SKIPPED: 10 | WARNINGS: 0 | FAILED: 2 ]
  1. Failure: distance matrix creation works (@test-distances.R#26) 
  2. Failure: get_local_distances works (@test-input_creation.R#134) 
  
  Error: testthat unit tests failed
  Execution halted**

Color palett of the ouput for color-blind people

The color scheme is very hard to see for people with color blindness especially red-green color blindness like I have. Would it be possible to add a selection argument for the plot function which would optionaly allow color-blind friendly schemes to be used?
The color options of this package https://www.fabiocrameri.ch/colourmaps/ would work very good for colorblind people.

rename get_geo_richness to get_richness or calculate_richness?

in the internal functions

CRAN complain for non-standard things in the check directory ... NOTE

* checking for non-standard things in the check directory ... NOTE
Found the following files/directories:
  'examples_i386' 'examples_x64' 'gen3sis-Ex_i386.Rout'
  'gen3sis-Ex_x64.Rout' 'tests_i386' 'tests_x64'

The above happens when checking the package.

plot_landscape function crash

returning:

Error in matrix(no_data, nrow = nrow(landscape[["coordinates"]]), ncol = ncol(values) +  : 
  non-numeric matrix extent
Called from: matrix(no_data, nrow = nrow(landscape[["coordinates"]]), 
    ncol = ncol(values) + 2, dimnames = list(rownames(landscape[["coordinates"]]), 
        c(colnames(landscape[["coordinates"]]), colnames(values))))
Error during wrapup: unimplemented type (29) in 'eval'

Error: no more error handlers available (recursive errors?); invoking 'abort' restart
Error during wrapup: INTEGER() can only be applied to a 'integer', not a 'unknown type #29'
Error: no more error handlers available (recursive errors?); invoking 'abort' restart

Once landscape input is only for a certain environment (e.g. landscape$temp) it returns:

plot_landscape(l$temp)
Error in matrix(no_data, nrow = nrow(landscape[["coordinates"]]), ncol = ncol(values) +  : 
  non-numeric matrix extent
Called from: matrix(no_data, nrow = nrow(landscape[["coordinates"]]), 
    ncol = ncol(values) + 2, dimnames = list(rownames(landscape[["coordinates"]]), 
        c(colnames(landscape[["coordinates"]]), colnames(values))))
Error during wrapup: unimplemented type (29) in 'eval'

Error: no more error handlers available (recursive errors?); invoking 'abort' restart
Error during wrapup: INTEGER() can only be applied to a 'integer', not a 'unknown type #29'
Error: no more error handlers available (recursive errors?); invoking 'abort' restart

changing landscape[["environment"]] does not help.

export and document get_geo_richness

so that users can have this handy function as well

plot_richness

plot_richness causes gen3sis to crash when called in the observer function when all species have gone extinct.

plot_richness(species, landscape)
Error in (mult * length(col)):1 : NA/NaN argument
In addition: Warning message:
In x[] <- col[x] :
number of items to replace is not a multiple of replacement length

The problem is there is no color in the richness palette for values of 0

max(richness)
[1] 0
color_richness(max(richness, na.rm=TRUE))
[1] character(0)
color_richness(1)
[1] "#440154"

This problem also means colours on a richness plot cannot distinguish between cells with 1 species and cells with 2 species.

Habitability mask during Input creation

Currently the habitability mask in the input creation overrides the habitable cells in the landscape rasters. This is going to be inconsistent with NA-cells in the input rasters as these are used to determine habitable cells during the simulation run.

I think a better/more intuitive way would be to use the habitability mask in addition to the landscapes, as such that for example it allows to limit the world to a smaller extract.

Check main page

fyi, I have activated the project page, which displays the content of the readme.md file here https://project-gen3sis.github.io/R-package/

The readme is at the moment not very helpful to the user. Would suggest to add info for how to install etc. similar to here http://florianhartig.github.io/DHARMa/ or whatever you think is most suitable.

add species ids as names in saving functions (besides save_species)

currently the species id and it's index in the species list match, nevertheless to it is ambiguous to implicitly rely on that for saved states (e.g save_traits)

versions history

releases summary files. maybe just a a single file e.g. releases_history.txt? any other idea?

config_handling.R line:122 requires a value!

Warning: @return [C:\VITAL LOCAL\Meus Documentos\ETH PhD\Code\R\package\Gen3sis\R\config_handling.R#122]: requires a value

relaunch simulations from val intermediate objects

Missing help file

Warning: [/Users/florian/Dropbox/Home/Projekte/Papers_in_Review/19-rGAsm-package/R-package/R/observations.R:146] @example '/Users/florian/Dropbox/Home/Projekte/Papers_in_Review/19-rGAsm-package/R-package/inst/examples/observer_plot_help.R' doesn't exist

-> should I add the file?

add colaborations file to git

* checking Rd files ... WARNING checkRd: (5) gen3sis.Rd:0-14: Must have a \description

• checking Rd files ... WARNING

checkRd: (5) gen3sis.Rd:0-14: Must have a \description

add rds data (non-RData, csv or txt) to data or vignette

https://stackoverflow.com/questions/36237991/r-can-i-use-rds-files-for-my-data-in-a-package

build package NOTE

Returning this NOTE when checked on CRAN:

• checking LazyData ... NOTE 'LazyData' is specified without a 'data' directory

Relates to the update on Fri, 23 Apr 2021:
https://developer.r-project.org/blosxom.cgi/R-devel/NEWS/2021/04/23#n2021-04-23

CRAN Build feedback:
package gen3sis_1.3.tar.gz does not pass the incoming checks automatically, please see the following pre-tests:
Windows: https://win-builder.r-project.org/incoming_pretest/gen3sis_1.3_20210630_120636/Windows/00check.log
Status: 1 NOTE
Debian: https://win-builder.r-project.org/incoming_pretest/gen3sis_1.3_20210630_120636/Debian/00check.log
Status: 1 NOTE

Related issue forum:
https://stackoverflow.com/questions/66860659/lazydata-is-specified-without-a-data-directory-error-when-submitting-r-packa

Example files for help

I added an example for examples in the help in 1db8808 - will have to populate this for all important functions

add new check functions?

Update automated checks with new functions provided by ZHU related to #58

create_input scaling bug/problem

create_input scales from meters to kilometers by multiplying a factor of 1/1000. This only works if a coordinate system is set for the input rasters.

line ~159 in Input_creation.R:
correction@transitionMatrix@x <- (1/1000) / correction@transitionMatrix@x

Should we remove that and tell the user the cost/distances are in meters, if they want kilometers they have to scale their cost function by 1/1000?

introduction vignette, richness plot

one of the richness plots is in the inst/logo folder, should be moved to extdata/southamerica/images

add travis win build

add windows and possibly mac flavours on CI travis

plot_summary fails if simulation ends before reaching step 0

make_summary hardcodes richness_final to
sgen3sis$summary <- c(data$summaries, list("richness-final"=data$geo_richness[,c(1,2,ncol(data$geo_richness))]))

This will be all NAs if the simulation aborts (too many species) or if end_time is not 0.
Subsequently this:

  ras <- rasterFromXYZ(output$summary$`richness-final`)
  rc <- color_richness(max(ras@data@values, na.rm=TRUE) + 1)

will call max() with an all NA vector which will return -Inf. Of course a colorRampPalette can't produce -Inf many colours and it fails.

path magic feature and bug

Currently there is quite some magic going on trying to divine the config, input, and output paths from partial info when running a simulation. Currently it is a bit inconsistent and intransparent on what is going on. Should we remove all/most of it and instead expect all three to be provided when calling a simulation?

The only part I'd keep is creating a subfolder with the config name in the output path.

(config_handling.R, prepare_directories, lines 16-81)

Automatic tests / continuous integration

Hi guys,

just as we talked about the tests - you haven't set up CI tests yet, right? There are at least two options, Travis CI, which I do for DHARMa, and there is a new option directly via GitHub workflows, see an example here https://github.com/TheoreticalEcology/s-jSDM/actions

The advantage of using the workflows is that a) I think it can also be done on a private repo b) it also tests Windows and MacOS. Advantage of TravisCI is that it looks nicer (in my opinion), especially when you work a lot with branches.

Anyway, we should decide for one or the other, this will help a lot with the tests and questions such as #6

misplaced loop_ecology() call

loop_ecology() is currently called in line 270 in gen3sis_main.R script. As the observer is invoked before, observer output does not reflect ecology processes of this time step.

R-package/R/gen3sis_main.R

Line 270 in d7a65d8

val <- loop_ecology(val$config, val$data, val$vars)

Documentation and comments in the code suggest that it should be called after the evolution step and prior to the end of loop updates in line 232.

R-package/R/gen3sis_main.R

Lines 226 to 232 in d7a65d8

	# #--------------------------------------------------------#
	# ######## loop ecology #######
	# #--------------------------------------------------------#
	if(verbose>=2){
	cat("ecology \n")
	}

add option for saving plots to files at the plot folder

vignette problems in CRAN Should be corrected by 2021-10-19 to safely retain gen3sis on CRAN.

Please see the problems shown on
https://cran.r-project.org/web/checks/check_results_gen3sis.html.

The vignette problems are from vignettes that contain unbalanced code chunk delimiters (e.g., a chunk opened by five backticks but closed by four, or opened by three but closed by four, etc.).

See yihui/knitr#2057 for more information.

Please correct before 2021-10-19 to safely retain your package on CRAN.