[Rd] reverse dependency checks

[Johannes Ranke]

Am Donnerstag, 18. Februar 2021, 19:58:09 CET schrieb Therneau, Terry M., 
Ph.D. via R-devel:
> There are some nice tools to automate reverse dependency checks, but for a
> large package the real issue is the envirionment.  The description of the
> crandalf site on github has a nice summary.  One package uses a cryto
> libraries (oops, install those), another uses some latex macros I've never
> heard of, etc.    In any case, I am seeing some failures in my reverse
> checks for survival that I'm not sure how to fix.   If anyone has a hint or
> two I'd be grateful.
> 
> 1. A half dozen fail with "graphics API mismatch"

Such error messages could occur if visual diffs (of SVG plots) are run with R > 
4.0 (i.e. R-devel) where the reference files were generated with R <= 4.0. This 
is described in

  https://github.com/r-lib/vdiffr/issues/86

However the vdiffr tests are only run if you set the NOT_CRAN environment 
variable, for example because you use devtools for running tests, see

  https://github.com/r-lib/vdiffr#testing-versus-monitoring

Kind regards,

Johannes

> 
> 2. Over a dozen have a latex failure of
> ! pdfTeX error (\pdfsetmatrix): Unrecognized format..
> \AtBegShi using Output ...ipout \box \AtBeginShipoutBox
>                                                    \fi \fi
> 
> A check of kpsewhich shows I have all the packages, and I don't see a
> pattern in the .tex files at the point of failure.
> 
> Terry T.