[R-pkg-devel] Help needed with CRAN check error under MacOS and old-rel

[Duncan Murdoch]

On 2024-06-30 5:17 a.m., Aphalo, Pedro J wrote:
> Hello,
> A few days ago I submitted to CRAN an updated version of package 'ggpp', it passed checks during submission and version 0.5.8 is now on CRAN. However I see errors for old-rel only for MacOS at https://cran.r-project.org/web/checks/check_results_ggpp.html and I am stuck as I do not have where to rerun this test.
> 
> The error itself is surprising as it seems to be caused by not finding df1, which is just a data frame defined in a code chunk in the same vignette, which uses 'pkgdown'. This error is limited to MacOS R old-rel but affects both arm64 and x86_64. (For either of these, I could not find a working setup for GitHub integration or at https://mac.r-project.org/macbuilder/submit.html).
> 
> The offending code is in nudge-examples.Rmd starting at line 250. This file was last modified six-months' ago, i.e., it is the same as in the previous CRAN version of 'ggpp'.
> 
> (I would rather track the origin of the problem than disable all three possibly affected code examples.)
> 
> Thanks in advance for any help,

I would guess the issue is that `df1` is created conditionally on 
`eval_ggrepel`, but evaluated unconditionally.  Perhaps on that old 
platform `eval_ggrepel` was `FALSE`, because of some issue installing 
the `ggrepel` package.

I tried to see if `ggrepel` had issues, but it was updated today, and no 
issues are shown:  so perhaps your package will work today.  Still, I'd 
fix the "create conditionally, evaluate unconditionally" bug.

Duncan Murdoch