[Rd] Possible Regression in setClassUnion between 3.5.0 and 3.6.0

[Ezra Tucker]

Hi Robert,

This looks like a bug to me (tested in R 3.6.2 on Windows), f(new("a")) 
should return "hi!". I'll add that this DOES work properly in 3.6.1 
which leads me to suspect this could be due to the subtle change in the 
way method dispatch was performed to fix a different bug, in 3.6.2. Can 
anybody else confirm that?


On 2/18/2020 9:32 AM, Robert Harlow wrote:
> I am trying to create a class union of class unions to facilitate method
> dispatch. When I execute code in the global environment, everything acts as
> expected, however when I put the same code in the context of a package,
> selectMethod can no longer find the correct method. This first block below
> puts the code in the context of a package:
>
> fn <- "codefile.R"
> writeLines(
>      c(
>          "setClass('x', slots = list(slot ='character'))",
>          "setClass('y', slots = list(slot ='character'))",
>          "setClass('a', slots = list(slot ='character'))",
>          "setClass('b', slots = list(slot ='character'))",
>          "setClassUnion('xy', c('x', 'y'))",
>          "setClassUnion('ab', c('a', 'b'))",
>          "setClassUnion('xyab', c('xy', 'ab'))",
>          "setGeneric('f', function(object, ...) standardGeneric('f'))",
>          "setMethod('f', 'xyab', function(object, ...) print('hi!'))"
>      ),
>      con = fn
> )
> package.skeleton(code_files = "codefile.R")
> system("rm -rf anRpackage/man")
> system("R CMD INSTALL anRpackage")
> library(anRpackage)
> ## fails in R 3.6.2, but works in R 3.5.0
> f(new("a"))
>
> Next, if a fresh R 3.6.2 session is started and I execute the following at
> the prompt, method dispatch works as expected.
>
> setClass("x", slots = list(slot ="character"))
> setClass("y", slots = list(slot ="character"))
> setClass("a", slots = list(slot ="character"))
> setClass("b", slots = list(slot ="character"))
> setClassUnion("xy", c("x", "y"))
> setClassUnion("ab", c("a", "b"))
> setClassUnion("xyab", c("xy", "ab"))
> setGeneric("f", function(object, ...) standardGeneric("f"))
> setMethod("f", "xyab", function(object, ...) print("hi!"))
> ## print's "hi!" as expected
> f(new("a"))
>
> I have also posted to stack overflow here:
> https://stackoverflow.com/questions/60264786/r-s4-class-union-of-class-unions?noredirect=1#comment106627883_60264786,
> (the example in this note removes devtools to make the environment
> cleaner).
>
> Interestingly, the issue only seems to arise when there are > 1 layer
> of the class union. E.g. if I were to setMethod on "ab" instead of
> "xyab", method dispatch would work as expected. I am not posting a bug
> yet as it is still unclear to me if I am doing something incorrect.
>
> My sessionInfo() is :
>
> R version 3.6.2 (2019-12-12)
> Platform: x86_64-pc-linux-gnu (64-bit)
> Running under: Ubuntu 18.04.3 LTS
>
> Matrix products: default
> BLAS:   /usr/local/lib/R/lib/libRblas.so
> LAPACK: /usr/local/lib/R/lib/libRlapack.so
>
> locale:
>   [1] LC_CTYPE=en_US.UTF-8       LC_NUMERIC=C
>   [3] LC_TIME=en_US.UTF-8        LC_COLLATE=en_US.UTF-8
>   [5] LC_MONETARY=en_US.UTF-8    LC_MESSAGES=en_US.UTF-8
>   [7] LC_PAPER=en_US.UTF-8       LC_NAME=C
>   [9] LC_ADDRESS=C               LC_TELEPHONE=C
> [11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C
>
> attached base packages:
> [1] stats     graphics  grDevices utils     datasets  methods   base
>
> other attached packages:
> [1] anRpackage_1.0
>
> loaded via a namespace (and not attached):
> [1] compiler_3.6.2 tools_3.6.2
>
> Thanks in advance for the help!
>
> 	[[alternative HTML version deleted]]
>
> ______________________________________________
> R-devel using r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel