Closed t-kalinowski closed 1 year ago
Looks like one small patch in the tools package will be necessary before S7 can pass R CMD check with the new, updated (and correct) matrixOps signature. https://github.com/r-devel/r-svn/pull/134/files. I believe the change is confined to the static-analysis checks only - my understanding is both Ops
and matrixOps
groups match arguments positionally only when evaluated normally, with argument names ignored.
I built R locally with the small patch applied and confirmed that S7 passes R CMD check.
Why are we seeing build failures on the latest release?
The {tools} package was enforcing the wrong signature for matrixOps methods. This is fixed and will be part of R 4.4.0 (afaik)( https://github.com/r-devel/r-svn/pull/134, https://github.com/r-devel/r-svn/commit/b0abc0edbe4ae1a61ee95bb1f049014b8738a2dd).
To avoid the R CMD check warning with the current release and older versions of R, I've added a special case that will define matrixOps.S7_object
with the "wrong" signature for R versions prior to 4.4.0.
Perfect, thanks!
This PR:
x, y
toe1, e2
, matching base R.matrixOps.S7_object
is no longer an alias ofOps.S7_object
, andOps.S7_object
no longer dispatches%*%
.closes #321