Dplyr: R CMD check problems

[krlmlr]

https://cran.r-project.org/web/checks/check_results_dplyr.html

Need to replicate locally. Also, the note on four UTF-8 marked strings in the data would spoil an overall "OK" result, maybe we can convert the data to ASCII.

[JohnMount]

The checks went red right after rlang 0.2.0 was pushed to CRAN https://github.com/r-lib/rlang/commit/01c57762b57e05fabde66c96798d05a275832efd .

[krlmlr]

Thanks. Now testing archive/0.7.4-current with RDsan of https://github.com/wch/r-debug.

[krlmlr]

Got nowhere with RDsan, retrying tomorrow.

[krlmlr]

@lionel-: If the CRAN failures started to occur with the CRAN release of rlang 0.2.0, could this be an ABI incompatibility of sorts? Have you tried installing rlang < 0.2.0, then dplyr, then rlang 0.2.0?

[krlmlr]

@lionel-: I was confused, I don't think an ABI change causes this, because the package under test is rebuilt with R CMD check. Will look further.

[krlmlr]

Fixed in 866df05.

[wlandau]

More CRAN check errors just popped up, and they all seem to relate to tidy eval. And it's not just dplyr. drake is having problems too (https://github.com/ropensci/drake/issues/356), so I am very interested in finding out what is happening.

[lionel-]

@wlandau Deparsing of !! is now fixed in R 3.5.0, i.e. !!x is now deparsed to "!!x" instead of "!(!x)". Using deparse makes unit tests dependent on the way R formats deparsed expressions and so may be brittle.

[wlandau]

So glad to hear that the deparsing of !! was fixed and that these errors are happening for a good reason.

[lock[bot]]

This old issue has been automatically locked. If you believe you have found a related problem, please file a new issue (with reprex) and link to this issue. https://reprex.tidyverse.org/