[Rd] Change to I() in R 4.1

[Martin Maechler]

> 
> Hi there,
> Is that change in R-devel intentional?
> 
>    library(Matrix)
>    m <- as(matrix(c(0, 1)), "sparseMatrix")
> 
>    isS4(m)
>    # [1] TRUE
> 
>    x <- I(m)
>    # Warning message:
>    # In `class<-`(x, unique.default(c("AsIs", oldClass(x)))) :
>    #   Setting class(x) to multiple strings ("AsIs", "dgCMatrix", ...); 
> result will no longer be an S4 object
> 
>    isS4(x)
>    # [1] FALSE
> 
> This works fine in R 4.0.3 i.e. no warning and I() doesn't turn off the 
> S4 bit of the object.
> 
> This change breaks 17 Bioconductor packages.
> 
> Seems that the culprit is this change in how I() is implemented:
> 
> In R 4.0.3:
> 
>    > I
>    function (x)
>    {
>      structure(x, class = unique(c("AsIs", oldClass(x))))
>    }
> 
> In R devel:
> 
>    > I
>    function (x)
>    `class<-`(x, unique.default(c("AsIs", oldClass(x))))

Yes, (by me),  as  I()  was sticking out in the slowness bug PR#17794
  https://bugs.r-project.org/bugzilla/show_bug.cgi?id=17794

and the direct dangerous  `call<-`  will be replaced happily in
I()'s definition.

*But* as Luke Tierney had remarked to R-core, direct changing of
the class of an S4 object has given the above warning for a
quite while, (svn r47934 | jmc | 2009-02-17 )
*and* it has rather been an inconsistency in R, that you could
still use "low-level" means to change the class of an S4 object
to something invalid.

I really don't think people should be allowed to use I() to
change a valid S4 object into an invalid one, but this is what
happens (R 4.0.3 patched) :

> require(Matrix); M <- Matrix(0, 2,3); IM <- I(M)
> validObject(IM)
Error in .classEnv(classDef) : 
  trying to get slot "package" from an object of a basic class ("NULL") with no slots
> IM
 ----------- FAILURE REPORT -------------- 
 --- failure: length > 1 in coercion to logical ---
 --- srcref --- 
: 
 --- package (from environment) --- 
methods
 --- call from context --- 
showDefault(object)
 --- call from argument --- 
!is.null(clDef) && isS4(object) && is.na(match(clDef using className, 
    .BasicClasses))
 --- R stacktrace ---
where 1: showDefault(object)
where 2: Error in showDefault(object) : 
  cannot get a slot ("slots") from an object of type "NULL"
> 


> Unfortunately there is a bunch of code around that calls I() on S4 
> objects, admittedly not necessarily for very good reasons, but it 
> happens. Would it be possible that I() has a less destructive effect on 
> S4 objects?

I'm not sure if this is really desirable... but I may fail to
see the point of allowing invalid   I(<S4>)  objects as they
appear in R 4.0.x  ..

So what do you really propose that  I(.) should be doing, e.g.,
for  'M'  above ?

Martin


> 
> Thanks,
> H.
> 
>  > sessionInfo()
> R Under development (unstable) (2020-10-17 r79346)
> Platform: x86_64-pc-linux-gnu (64-bit)
> Running under: Ubuntu 20.04.1 LTS
> 
> Matrix products: default
> BLAS:   /home/biocbuild/bbs-3.13-bioc/R/lib/libRblas.so
> LAPACK: /home/biocbuild/bbs-3.13-bioc/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] Matrix_1.2-18
> 
> loaded via a namespace (and not attached):
> [1] compiler_4.1.0  grid_4.1.0      lattice_0.20-41
> 
> -- 
> Hervé Pagès
> 
> Program in Computational Biology
> Division of Public Health Sciences
> Fred Hutchinson Cancer Research Center
> 1100 Fairview Ave. N, M1-B514
> P.O. Box 19024
> Seattle, WA 98109-1024
> 
> E-mail: hpages using fredhutch.org
> Phone:  (206) 667-5791
> Fax:    (206) 667-1319
> ______________________________________________
> R-devel using r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel