isIncluded for all data types?

[PavelBal]

Following from the documentation of the function isIncluded it is not clear to me, which data types are supported. Probably only atomic, and it cannot check whether e.g. a certain DataSet is inside a list of DataSet? In this case, it should be documented accordingly.

[PavelBal]

> isIncluded(c(1,2,3), list(c(1,2,3)))
[1] FALSE

So definitely only for atomics...

[PavelBal]

> isIncluded(1, "1")
[1] TRUE

(veryconfused) ok lol, I don't like this function :D

[IndrajeetPatil]

This behavior actually has nothing to do with this function and everything to do with implicit and arcane coercion rules of R

1 %in% "1"
#> [1] TRUE

^{Created on 2022-02-02 by the reprex package (v2.0.1.9000)}

[msevestre]

Lol. Arcane coercion rules..

I am going to name this the truth of the month. Respect @IndrajeetPatil

[IndrajeetPatil]

@PavelBal Is this what you want to do?

it cannot check whether e.g. a certain DataSet is inside a list of DataSet

> library(ospsuite)
> library(ospsuite.utils)
> 
> # import observed data
> dataSet <- loadDataSetsFromExcel(
+   xlsFilePath = file.path(getwd(), "tests", "data", "CompiledDataSetStevens2012.xlsx"),
+   importerConfiguration = DataImporterConfiguration$new(file.path(getwd(), "tests", "data", "ImporterConfiguration.xml"))
+ )
> 
> isIncluded("Stevens_2012_placebo.Placebo_total", names(dataSet))
[1] TRUE

[PavelBal]

Is this what you want to do?

No, I meant that it must be documented that it only works on atomic data types. E.g. this would probably not work:

ds1 <- DataSet$new("DataSet1")
ds2 <- DataSet$new("DataSet1")
ds3 <- DataSet$new("DataSet1")

isIncluded(ds1, list(ds1, ds2, ds3))

And as already shown before - it does not work on lists:

> isIncluded(c(1,2,3), list(c(1,2,3)))
[1] FALSE

[IndrajeetPatil]

it only works on atomic data types

The function works with a vector of values, be it an atomic or non-atomic vector. So all you need to do for this to work with DataSet objects, is to convert the argument to a vector.

library(ospsuite)
#> Loading required package: rClr
#> Loading the dynamic library for Microsoft .NET runtime...
#> Loaded Common Language Runtime version 4.0.30319.42000
library(ospsuite.utils)

ds1 <- DataSet$new(name = "DataSet1")
ds2 <- DataSet$new(name = "DataSet1")
ds3 <- DataSet$new(name = "DataSet1")

is.vector(ds1)
#> [1] FALSE
is.vector(c(ds1))
#> [1] TRUE

isIncluded(ds1, list(ds1, ds2, ds3))
#> Error in match(x, table, nomatch = 0L): 'match' requires vector arguments
isIncluded(c(ds1), list(ds1, ds2, ds3))
#> [1] TRUE

^{Created on 2022-02-02 by the reprex package (v2.0.1.9000)}

it does not work on lists

It works with lists, and it's telling you correctly that the first vector is not contained in the second list. In fact, they are not even of the same length: vector has a length of 3, while list has a length of 1. So the function is working as expected.

library(ospsuite.utils)

length(c(1, 2, 3))
#> [1] 3
length(list(c(1, 2, 3)))
#> [1] 1

isIncluded(c(1, 2, 3), list(c(1, 2, 3)))
#> [1] FALSE

isIncluded(c(1, 2, 3), list(1, 2, 3))
#> [1] TRUE
isIncluded(list(1, 2, 3), c(1, 2, 3))
#> [1] TRUE

^{Created on 2022-02-02 by the reprex package (v2.0.1.9000)}

[PavelBal]

isIncluded(ds1, list(ds1, ds2, ds3))
#> Error in match(x, table, nomatch = 0L): 'match' requires vector arguments
isIncluded(c(ds1), list(ds1, ds2, ds3))
#> [1] TRUE

Sorry, but this just does not work. You cannot use e.g. R6 objects with this function.

> isIncluded(c(ds1), list(ds2))
[1] TRUE

> isIncluded(c(ds1), list(DataSet))
[1] TRUE

I am ok with this as long it is clear to the user ( == documented properly).

[PavelBal]

Re-opening it because the behavior is currently erroneous regarding non-primitive types

So this function only works for primitive types basically.
Does it make sense to say that we throw an error if it is used with something else than char, int, double, boolean etc..?

Yes I would vote for this. This function has (hopefully) not been used with R6 classes or whatever, so let's not invest time unless we have a need for it.

BTW yes, identical is cool, I think we all should read the documentation for it and consider it in our code (especially in if statements).

> NA == NA
[1] NA
> identical(NA, NA)
[1] TRUE

Hell yeah, THIS makes sense!

[PavelBal]

@msevestre @IndrajeetPatil @pchelle

[IndrajeetPatil]

Haha

There is an entire book on the weird behaviour R sometimes exhibit:

[IndrajeetPatil]

To be fair to R though, it's not unique in this regard. Watch the notorious WAT talk for a similar take on JavaScript:

https://www.youtube.com/watch?v=et8xNAc2ic8